<<

Land-Based Wind Market Report: 2021 Edition This report is being disseminated by the U.S. Department of (DOE). As such, this document was prepared in compliance with Section 515 of the Treasury and General Government Appropriations Act for fiscal year 2001 (public law 106-554) and information quality guidelines issued by DOE. Though this report does not constitute “influential” information, as that term is defined in DOE’s information quality guidelines or the Office of Management and Budget’s Information Quality Bulletin for Peer Review, the study was reviewed both internally and externally prior to publication. For purposes of external review, the study benefited from the advice and comments of 11 industry stakeholders, U.S. Government employees, and national laboratory staff.

NOTICE This report was prepared as an account of work sponsored by an agency of the United States government. Neither the United States government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States government or any agency thereof.

Available electronically at SciTech Connect: http://www.osti.gov/scitech

Available for a processing fee to U.S. Department of Energy and its contractors, in paper, from: U.S. Department of Energy Office of Scientific and Technical Information P.O. Box 62 Oak Ridge, TN 37831-0062 OSTI: http://www.osti.gov Phone: 865.576.8401 Fax: 865.576.5728 Email: [email protected]

Available for sale to the public, in paper, from: U.S. Department of Commerce National Technical Information Service 5301 Shawnee Road Alexandria, VA 22312 NTIS: http://www.ntis.gov Phone: 800.553.6847 or 703.605.6000 Fax: 703.605.6900 Email: [email protected]

ii Land-Based Wind Market Report

Preparation and Authorship This report was prepared by Lawrence Berkeley National Laboratory for the Wind Energy Technologies Office of the U.S. Department of Energy’s Office of Energy Efficiency and .

Corresponding authors of the report are: Ryan Wiser and Mark Bolinger, Lawrence Berkeley National Laboratory. The full author list includes: Ryan Wiser, Mark Bolinger, Ben Hoen, Dev Millstein, Joe Rand, Galen Barbose, Naïm Darghouth, Will Gorman, Seongeun Jeong, Andrew Mills, and Ben Paulos.

iii Land-Based Wind Market Report

Acknowledgments For their support of this ongoing report series, the authors thank the entire U.S. Department of Energy (DOE) Wind Energy Technologies Office team. In particular, we acknowledge Patrick Gilman and Robert Marlay. For reviewing elements of this report or providing key input, we also thank: Christopher Namovicz, Manussawee Sukunta, and Richard Bowers (U.S. Energy Information Administration); Andrew David (U.S. International Trade Commission); Charlie Smith (Energy Systems Integration Group); Feng Zhao (Global Wind Energy Council); David Milborrow (consultant); Anjelaka Stolte (Boston Consulting Group); Matt McCabe (Clear Wind); Eric Lantz (National Renewable Energy Laboratory, NREL); Kelsey Bartz, Brendan Casey, and John Hensley (American Clean Power Association); and Patrick Gilman and Gage Reber (DOE). For providing data that underlie aspects of the report, we thank the U.S. Energy Information Administration, BloombergNEF, Wood Mackenzie, Global Wind Energy Council, and the American Clean Power Association. Thanks also to Donna Heimiller (NREL) for assistance in mapping wind resource quality; Amy Howerton and Carol Laurie (NREL), and Liz Hartman, Coryne Tasca, and Kaitlyn Roach (DOE) for assistance with layout, formatting, production, and communications. Lawrence Berkeley National Laboratory’s contributions to this report were funded by the Wind Energy Technologies Office, Office of Energy Efficiency and Renewable Energy of the DOE under Contract No. DE-AC02-05CH11231. The authors are solely responsible for any omissions or errors contained herein.

iv Land-Based Wind Market Report

List of Acronyms ACP American Clean Power Association BNEF BloombergNEF BPA Bonneville Power Administration CAISO California Independent System Operator COD commercial operation date CCA community choice aggregator DOE U.S. Department of Energy EIA U.S. Energy Information Administration ERCOT Electric Reliability Council of Texas FAA Federal Aviation Administration FERC Federal Energy Regulatory Commission GE General Electric Corporation GW gigawatt HTS Harmonized Tariff Schedule IOU investor-owned utility IPP independent power producer ISO independent system operator ISO-NE New England Independent System Operator ITC investment tax credit kV kilovolt kW kilowatt kWh kilowatt-hour LCOE levelized cost of energy m2 square meter MISO Midcontinent Independent System Operator MW megawatt MWh megawatt-hour NREL National Renewable Energy Laboratory NYISO New York Independent System Operator O&M operations and maintenance OEM original equipment manufacturer PJM PJM Interconnection POU publicly owned utility PPA power purchase agreement PTC production tax credit

v Land-Based Wind Market Report

REC renewable energy certificate RPS renewables portfolio standard RTO regional transmission organization SGRE Siemens Gamesa Renewable Energy SPP Southwest Power Pool W WAPA Western Area Power Administration WECC Western Electricity Coordinating Council

vi Land-Based Wind Market Report

Executive Summary capacity additions in the United States hit a new record in 2020, supported by the industry’s primary federal incentive—the production tax credit (PTC)—as well as a myriad of state-level policies. Improvements in the cost and performance of wind power technologies have also driven wind capacity additions, yielding low-priced wind energy for utility, corporate, and other power purchasers.

Key findings from this year’s Land-Based Wind Market Report—which primarily focuses on land-based, utility-scale wind—include: Installation Trends • Wind power capacity grew at a record pace in 2020, with 16,836 MW of new capacity added in the United States and $24.6 billion invested. Cumulative wind capacity grew to 121,985 megawatts (MW). In addition, 3,087 MW of existing wind were partially repowered in 2020, mostly by upgrading rotors and major nacelle components of existing turbines. • Wind power represented the largest source of U.S. electric-generating capacity additions in 2020. Wind power constituted 42% of all capacity additions in 2020. Over the last decade, wind represented 29% of total U.S. capacity additions, and an even larger fraction of new capacity in SPP (75%), ERCOT (54%), MISO (52%), and the non-ISO West (32%). [See Figure 1 for regional definitions]. • Globally, the United States ranked second in annual wind capacity additions in 2020, but remained well behind the market leaders in wind energy penetration. Global wind additions hit a record in 2020, with nearly 93 GW of newly added capacity, yielding a cumulative total 743 GW. The United States remained the second-leading market in terms of annual and cumulative capacity, behind China. A number of countries have achieved high levels of wind penetration, with wind supplying nearly 50% of Denmark’s total in 2020, and between 25% and 40% in Ireland, Germany, the U.K, and Portugal. In the United States, wind supplied 8.3% of total electricity generation in 2020. • Texas installed the most capacity in 2020 with 4,137 MW, while sixteen states exceeded 10% wind energy penetration as a fraction of total in-state generation. Texas also remained the clear leader on a cumulative basis, with 32,686 MW of capacity. Notably, the wind capacity installed in Iowa supplied 57% of all in-state electricity generation in 2020, while Kansas (43%), Oklahoma (35%), South Dakota (33%) and North Dakota (31%) were all above 30% by this metric. Within independent system operators (ISOs), 2020 wind penetration (expressed as a percentage of load) was 31.3% in SPP, 22.7% in ERCOT, 11.0% in MISO, 6.6% CAISO, 3.4% in the PJM, 3.0% in ISO-NE, and 2.9% in NYISO. • A small but growing number of hybrid plants that pair wind with storage and other resources are operating in the United States. There were 38 hybrid wind power plants in operation at the end of 2020, representing 2.3 GW of wind and 0.9 GW of co-located assets. The most common wind hybrid project combines wind and storage technology, where 1.4 GW of wind has been paired with 0.2 GW of battery storage (14% storage to generator ratio). The average storage duration of these projects is 0.6 hours, suggesting a focus on ancillary services and limited capacity to shift large amounts of energy across . • Despite a slight contraction since 2018, substantial wind power capacity exists in transmission interconnection queues; solar and storage reached new highs in 2020. At the end of 2020, there were 209 gigawatts (GW) of wind capacity seeking transmission interconnection, including 61 GW of offshore wind. In 2020, 55 GW of wind capacity entered interconnection queues, 10 GW of which are proposed as hybrid configurations. Energy storage interconnection requests have increased in recent years, both for stand-alone and hybrid plants, most-often pairing solar with storage. The SPP, West (non-ISO), and NYISO regions had the greatest quantity of wind in their queues at the end of 2020. Nearly half of all wind capacity added to interconnection queues in 2020 was for offshore wind plants.

vii Land-Based Wind Market Report

Industry Trends • GE and Vestas supplied turbines for 87% of U.S. wind power capacity installed in 2020. In 2020, GE captured 53% of the U.S. market for turbine installations, followed by Vestas at 34% and Siemens- Gamesa Renewable Energy (SGRE) at 9%, Nordex at 3%, and Goldwind with 1%. • The domestic wind industry supply chain was reasonably stable in 2020. Despite COVID-19, wind- related job totals in the United States increased in 2020, to 116,800 full-time workers. Domestic nacelle assembly capability remained at 15 GW in 2020, and the United States had the capability to produce blades and towers sufficient for approximately 9.4 GW and 10.3 GW, respectively, of wind capacity annually. Fierce competition has reduced turbine manufacturer profitability over the last several years. • Domestic manufacturing content is strong for some components, but the U.S. wind industry remains reliant on imports. The United States imports wind equipment from a wide array of countries, including most prominently in 2020: India, Spain, China, and Mexico. Domestic content is highest for nacelle assembly (>85%), towers (60%–75%), and blades and hubs (30%–50%), though domestic content for blades in particular has declined in recent years. • Project finance was volatile in 2020. Debt interest rates weathered pandemic-induced volatility in 2020, but ended the year only slightly lower than where they began. A record tax equity investment of ~$11 billion in wind was achieved in 2020. Nonetheless, tax equity grew more scarce throughout the year, resulting in higher yields by year’s end. Extensions of the PTC’s construction start deadline and safe harbor window in which to place a project in service aimed to ease the tax equity crunch. • Independent power producers own the majority of wind assets built in 2020. Independent power producers (IPPs) own 73% of the new wind capacity installed in the United States in 2020, with the remaining assets (27%) owned by investor-owned utilities. • Long-term contracted sales to utilities remained the most common offtake arrangement, but direct retail sales and merchant offtake arrangements were both significant. Electric utilities continued to be the largest offtakers of wind power in 2020, either owning wind projects (27%) or buying electricity from projects (32%) that, in total, represent 59% of the new capacity installed in 2020. Direct retail purchasers—including corporate offtakers—account for 27%. Merchant/quasi-merchant projects (12%) and power marketers (2%) make up the remainder. Technology Trends • Turbine capacity, rotor diameter, and hub height have all increased significantly over the long term. To optimize wind project cost and performance, turbines continue to grow in size. The average rated (nameplate) capacity of newly installed wind turbines in the United States in 2020 was 2.75 MW, up 8% from the previous year and 284% since 1998−1999. The average rotor diameter of newly installed turbines in 2020 was 124.8 meters, a 3% increase over 2019 and 159% over 1998−1999, while the average hub height was 90.1 meters, unchanged from the previous year and up 59% since 1998−1999. • Turbines originally designed for lower wind sites dominate the market. With growth in swept rotor area outpacing growth in nameplate capacity, there has been a decline in the average “specific power” 1 (in W/m2), from 393 W/m2 among projects installed in 1998–1999 to 223 W/m2 among projects installed in 2020. Turbines with low specific power were originally designed for lower sites. • Wind turbines were deployed in somewhat lower wind-speed sites in 2020 than in the previous seven years. Wind turbines installed in 2020 were located in sites with an average estimated long-term wind speed of 7.7 meters per second at a height of 80 meters above the ground—this is the lowest average long-term wind speed among newly built projects in the last eight years. Federal Aviation

1 A wind turbine’s specific power is the ratio of its nameplate capacity rating to its rotor-swept area. All else equal, a decline in specific power should lead to an increase in .

viii Land-Based Wind Market Report

Administration (FAA) and industry data on projects that are either “under construction” or in “advanced development” suggest that the sites likely to be built out over the next few years will, on average, have even lower average wind . Increasing hub heights help to partially offset these trends, enabling turbines to access higher wind speeds. • Low-specific-power turbines are deployed on a widespread basis; taller towers are seeing increased use in a wider variety of sites. Low specific power turbines continue to be deployed in all regions, and at both lower and higher wind speed sites. The tallest towers (i.e., those above 100 meters) are found in greater relative frequency in the upper Midwest and Northeastern regions and at lower wind speed sites. • Wind projects planned for the near future are poised to continue the trend of ever-taller turbines. The average “tip height” (from ground to blade tip extended directly overhead) among projects that came online in 2020 is 525 feet (160 meters), and FAA data suggest that future projects, including those under construction and in advanced development, will deploy even taller turbines. Among “proposed” turbines in the FAA permitting process, the average tip height reaches more than 660 feet (200 meters). • In 2020, thirty-three wind projects were partially repowered, most of which now feature significantly larger rotors and lower specific power ratings. These projects totaled 3,087 MW prior to repowering. Of the changes made to the turbines, larger rotors dominated, increasing the average rotor diameter by 14 meters, while reducing specific power by 25%, from 334 to 251 W/m2. The primary motivations for partial repowering have been to re-qualify for the PTC, while at the same time increasing energy production and extending the useful of the projects. Performance Trends • The average capacity factor in 2020 exceeded 40% among wind projects built in recent years, and reached 36% on a fleet-wide basis. The average 2020 capacity factor among projects built from 2014 to 2019 was 41.4%, compared to an average of 29.0% among projects built from 2004 to 2011, and 25.2% among projects built from 1998 to 2001. This improvement among more-recently built projects has pushed the cumulative fleet-wide capacity factor higher over time, reaching 36% in 2020. • State and regional variations in capacity factors reflect the strength of the wind resource. Based on projects built from 2015–2019, average capacity factors in 2020 were highest in central states and lower closer to the coasts. Not surprisingly, the state and regional rankings are roughly consistent with the relative quality of the wind resource in each region. • Turbine design and site characteristics influence performance, with declining specific power leading to sizable increases in capacity factor over the long term. The decline in specific power has been a major contributor to higher capacity factors, but has been offset in part by a tendency toward building projects at sites with lower annual average wind speeds. • Wind curtailment can differentially impact project performance across sites and regions. Across all independent system operators (ISOs), wind energy curtailment in 2020 stood at 3.4%—generally rising over the last five years. This average masks variation across regions and projects. MISO and ERCOT, for example, experienced average wind curtailment rates in 2020 of 5.0% and 4.6%, respectively. • Temporal variations in wind speed also impact performance. The strength of the wind resource varies from year to year; moreover, the degree of inter-annual variation differs from site to site (and, hence, also region to region). This temporal and spatial variation impacts project performance from year to year. In 2020, the index was slightly above its long-term average, driven in part by higher wind speeds in the non-ISO West region. • Wind project performance degradation also explains why older projects did not perform as well in 2020. Capacity factor data suggest some amount of performance decline with project age, though perhaps only once projects age beyond 10 years. The apparent decline in capacity factors as projects progress into their second decade partially explains why older projects—e.g., those built from 1998 to 2001—did not perform as well as newer projects in 2020.

ix Land-Based Wind Market Report

Cost Trends • Wind turbine prices remained well below levels seen a decade ago. After hitting an initial low of roughly $900 per kilowatt (kW) from 2000 to 2002, average turbine prices increased to $1,800/kW by 2008.2 Since then, wind turbine prices have steeply declined, even as performance has improved. Recent data suggest pricing most-typically in the $770–$850/kW range. These price reductions, coupled with improved turbine technology, have exerted downward on project costs and enabled reductions in wind power prices and the levelized cost of wind energy. • Lower turbine prices have driven reductions in reported installed project costs. The capacity- weighted average installed project cost within the 2020 sample stood at $1,460/kW. This is a decrease of more than 40% from the peak in average costs in 2009 and 2010, but is roughly on par with the costs experienced in the early 2000s—albeit with much larger turbines and improved performance today. Costs have largely held steady over the last three years. • Installed costs differed by region. ERCOT and the (non-California) Western states hosted the lowest- cost projects built in 2020, with average costs of $1,280/kW and $1,380/kW respectively. Higher average costs were experienced in other regions for projects installed in 2020. • Installed costs differed by project size. Installed costs for plants built in recent years exhibit economies of scale, with costs declining as project capacity increases. • Operations and maintenance costs varied by project age and commercial operations date. Despite limited data availability, projects installed over the past 15 years have, on average, incurred lower operations and maintenance (O&M) costs than older projects in their first years of operation. The data also suggest that O&M costs tend to increase as projects age, at least for the older projects in the sample. Power Sales Price and Levelized Cost Trends • Wind power purchase agreement prices remain low. After topping out above $70/MWh for PPAs executed in 2009, the national average levelized price of wind PPAs has dropped. In the Central region of the country, recent pricing is around or below $20/MWh. In the West and East, prices tend to average roughly $30/MWh. Low PPA prices have been facilitated by the combination of higher capacity factors, declining installed costs and operating costs, and low interest rates; the PTC has also been a key enabler. • Recent wind power purchase agreements are priced in the mid-teens in some cases. Focusing on the 135 PPAs in the sample executed since the start of 2014, levelized prices range from $8.6/MWh to $82.7/MWh, with many PPAs—particularly in the Central and West regions—priced in the mid-teens. Contract terms range from 10 to 35 years, with an average of 20 years. • LevelTen Energy’s PPA price indices confirm low but rising PPA prices, and regional variations. In contrast to the prices summarized above, which principally involve utility purchasers, LevelTen provides an index of PPA offers made to large, end-use customers. These data show that prices have generally risen over the last six quarters, and vary by ISO. Among regions reporting data, CAISO features the highest pricing (~$50/MWh); the lowest prices are found in ERCOT and SPP (~$20/MWh). • The (unsubsidized) average levelized cost of wind energy has fallen to around $33/MWh. Trends in the levelized cost of energy (LCOE) closely follow the PPA trends—i.e., generally decreasing from 1998 to 2005, rising through 2009, and then declining to 2020, albeit with a plateau over the last several years. The national average LCOE of wind project built in 2020—excluding the PTC—was $33/MWh. • Levelized costs vary by region, with the lowest costs in ERCOT, SPP, and the non-ISO West. The lowest LCOEs for projects constructed in 2020—only considering regions with a larger sample—are found in ERCOT ($29/MWh), SPP ($32/MWh), non-ISO West ($32/MWh), and MISO ($35/MWh).

2 All cost figures presented in the report are denominated in real 2020 dollars.

x Land-Based Wind Market Report

Cost and Value Comparisons • Despite low PPA prices, wind faces competition from solar and . The once-wide gap between wind and solar PPA prices has narrowed considerably in recent years, as solar prices have fallen more rapidly than wind prices. With the support of federal tax incentives, both wind and solar PPA prices are now below the projected cost of burning natural gas in existing gas-fired combined cycle units. • Wind PPA prices have been broadly attractive compared to wind’s grid-system value in wholesale power markets. Following the sharp drop in wholesale electricity prices (and, hence, wind energy market value) in 2009, average wind PPA prices tended to exceed the wholesale market value of wind through 2012. Continued declines in wind PPA prices brought those prices back in line with the market value of wind in 2013, and wind has generally remained competitive in subsequent years. The year 2020 was unique due to the pandemic, with low electricity demand in the spring and low natural gas prices leading to low electricity prices. The market value of wind in 2020 was the lowest in ERCOT and MISO, at $11/MWh, whereas the highest-value market was CAISO at $29/MWh. • The grid-system market value of wind varies by project location. The 10th to 90th percentile range in 2020 market value across wind projects located in ISOs spans $7/MWh to $29/MWh. Within a region, transmission congestion can noticeably reduce the grid-value of wind plants. In some situations, wind patterns are locally differentiated, and can lead to value enhancements versus plants located elsewhere. • The grid-system market value of wind tends to decline with wind penetration, impacted by generation profile, transmission congestion, and curtailment. The regions with the highest wind penetrations (SPP at 31%, ERCOT at 23%, and MISO at 11%) have generally experienced the largest reduction in wind’s value relative to average wholesale prices. In 2020, wind’s value was ~50% lower than average wholesale prices in SPP, ERCOT and MISO, ~40% lower in NYISO, and ~20% lower in CAISO, ISO-NE, and PJM. These value reductions were primarily caused by a combination of transmission congestion and wind generation profiles that were negatively correlated with wholesale prices. Curtailment had only a minimal impact. • The health and benefits of wind are larger than its grid-system value, and the combination of all three far exceeds the levelized cost of wind. Wind reduces emissions of carbon dioxide, nitrogen oxides, and sulfur dioxide, providing and climate benefits. Nationally, these benefits averaged $76/MWh-wind in 2020. Benefits were largest, ranging from $80/MWh to $120/MWh, in the Central, Midwest, and Mid-Atlantic regions. Values were lowest in New York ($31/MWh) and New England ($27/MWh). Focusing on wind plants built in 2020, the average climate, health, and grid-system value sums to almost three the average LCOE. Climate, health, and grid value averaged $46/MWh, $31/MWh and $15/MWh, respectively, compared to an average LCOE of $33/MWh. Future Outlook Energy analysts project that wind capacity additions will continue at a rapid clip in 2021 before declining for a couple years and then rebounding. Specifically, expected additions range from 13 GW to 16 GW in 2021. Forecasts for 2022 and 2023 show a downturn, before rebounding to 11−13 GW in 2024 and 2025. These expected trends are driven in part by the scheduled expiration of the federal PTC, and by anticipated growth in offshore wind in the mid-2020s. Near-term additions are also influenced by improvements in the cost and performance of wind power technologies, which contribute to low power sales prices, and corporate wind energy purchases and state-level renewable energy policies. Expectations for continued low natural gas prices put a damper on growth expectations, as do limited transmission infrastructure and competition from solar. Longer term, the prospects for wind energy will be influenced by the sector’s ability to continue to improve its economic position even in the face of competition from solar and natural gas. Corporate demand for clean energy and state-level policies will also continue to impact wind deployment, as will the buildout of transmission infrastructure. Finally, there have been recent proposals for a long-term extension of the PTC, a federal clean energy standard, and other national policies to support a clean energy transition. The fate of these proposals may determine the sector’s upside potential to exceed the projections summarized above.

xi Land-Based Wind Market Report

Table of Contents Executive Summary ...... vii 1 Introduction ...... 1 2 Installation Trends ...... 3 3 Industry Trends ...... 13 4 Technology Trends ...... 23 5 Performance Trends ...... 31 6 Cost Trends ...... 38 7 Power Sales Price and Levelized Cost Trends ...... 44 8 Cost and Value Comparisons ...... 50 9 Future Outlook ...... 62 References ...... 63 Appendix: Sources of Data Presented in this Report ...... 65

xii Land-Based Wind Market Report

List of Figures Figure 1. Regional boundaries overlaid on a map of average annual wind speed at 80 meters ...... 2 Figure 2. Annual and cumulative growth in U.S. wind power capacity ...... 3 Figure 3. Relative contribution of generation types in annual capacity additions ...... 4 Figure 4. Generation capacity additions by region over last ten years ...... 5 Figure 5. Wind energy penetration in subset of top global wind markets ...... 6 Figure 6. Location of wind power development in the United States ...... 7 Figure 7. Location and capacity of hybrid wind plants in the United States ...... 9 Figure 8. Design characteristics of hybrid power plants operating in the United States, for a subset of configurations ...... 9 Figure 9. Generation capacity in 42 selected interconnection queues from 2015 to 2020, by resource type ...... 10 Figure 10. Wind power capacity in 42 selected interconnection queues, by region ...... 11 Figure 11. Hybrid power plants in 42 selected interconnection queues, including wind and solar hybrid configurations ...... 11 Figure 12. Hybrid wind power plants in 42 selected interconnection queues at the end of 2020 ... 12 Figure 13. Annual U.S. market share of wind turbine manufacturers by MW, 2005–2020 ...... 13 Figure 14. Location of turbine and component manufacturing facilities ...... 14 Figure 15. Domestic wind manufacturing capability vs. U.S. wind power capacity installations ...... 15 Figure 16. Turbine OEM global profitability ...... 15 Figure 17. Imports of wind-related equipment that can tracked with trade codes ...... 16 Figure 18. Summary map of tracked wind-specific imports in 2020: top-10 countries of origin and states of entry ...... 17 Figure 19. Origins of U.S. imports of selected wind turbine equipment in 2020 ...... 18 Figure 20. Approximate domestic content of major components in 2020 ...... 19 Figure 21. Cost of 15-year debt and tax equity for wind projects over time ...... 20 Figure 22. Cumulative and 2020 wind power capacity categorized by owner type ...... 21 Figure 23. Cumulative and 2020 wind power capacity categorized by power offtake arrangement 22 Figure 24. Average turbine nameplate capacity, hub height, and rotor diameter for land-based wind projects ...... 23 Figure 25. Trends in turbine nameplate capacity, hub height, and rotor diameter ...... 24 Figure 26. Trends in turbine specific power ...... 24

xiii Land-Based Wind Market Report

Figure 27. Wind resource quality by year of installation at 80 meters ...... 26 Figure 28: Location of low specific power turbine installations ...... 27 Figure 29: Location of tall tower turbine installations ...... 27 Figure 30. Total turbine heights proposed in FAA applications, over time ...... 28 Figure 31. Total turbine heights proposed in FAA applications, by location ...... 29 Figure 32. Change in average physical specifications of turbines that were partially repowered in 2020 ...... 30 Figure 33. Calendar year 2020 capacity factors by commercial operation date ...... 32 Figure 34. Average calendar year 2020 capacity factor by state ...... 32 Figure 35. 2020 capacity factors and various drivers by commercial operation date ...... 33 Figure 36. Calendar year 2020 capacity factors by wind resource quality and specific power: 1998- 2019 projects ...... 34 Figure 37. Wind curtailment and penetration rates by ISO ...... 35 Figure 38. Inter-annual variability in the wind resource by region and nationally ...... 36 Figure 39. Changes in project-level capacity factors as projects age ...... 37 Figure 40. Reported wind turbine transaction prices over time ...... 38 Figure 41. Installed wind power project costs over time ...... 39 Figure 42. Installed wind power project costs by region, over time ...... 40 Figure 43. Installed wind power project costs by region, in 2020 ...... 40 Figure 44. Installed wind power project costs by project size: 2019 and 2020 projects ...... 41 Figure 45. Average O&M costs for available data years from 2000 to 2020, by commercial operation date ...... 42 Figure 46. Median annual O&M costs by project age and commercial operation date ...... 43 Figure 47. Levelized wind PPA prices by PPA execution date and region (full sample)...... 45 Figure 48. Generation-weighted average levelized wind PPA prices by PPA execution date and region ...... 45 Figure 49. Levelized wind PPA prices by PPA execution date and region (recent sample) ...... 46 Figure 50. LevelTen Energy wind PPA price index by quarter of offer ...... 47 Figure 51. Estimated levelized cost of wind energy by commercial operation date ...... 48 Figure 52. Estimated levelized cost of wind energy, by region ...... 48 Figure 53. Levelized wind and solar PPA prices and levelized gas price projections ...... 50 Figure 54. Wind PPA prices and natural gas fuel cost projections by calendar year over time...... 51

xiv Land-Based Wind Market Report

Figure 55. Regional wholesale market value of wind and average levelized long-term wind PPA prices over time ...... 53 Figure 56. Regional wholesale market value of wind in 2020, by region ...... 54 Figure 57. Project-level wholesale market value of wind in 2020 ...... 55 Figure 58. Trends in wind value factor as wind penetrations increase ...... 56 Figure 59. Impact of transmission congestion, output profile, and curtailment on wind energy market value in 2020 ...... 57 Figure 60. Marginal health and climate benefits from wind generation by region in 2020 ...... 58 Figure 61. Marginal health, climate and grid-value benefits from new wind plants versus LCOE in 2020 ...... 58 Figure 62. Wind power capacity additions: historical installations and projected growth...... 62

List of Tables Table 1. International Rankings of Wind Power Capacity: Land-Based and Offshore ...... 6 Table 2. U.S. Wind Power Rankings: The Top 20 States ...... 8 Table A1. Harmonized Tariff Schedule (HTS) Codes and Categories Used in Wind Import Analysis . 66

xv

1 Introduction Wind power capacity additions in the United States hit a record high in 2020. Recent and projected near-term growth is supported by the industry’s primary federal incentive—the production tax credit (PTC)—as well as a myriad of state-level policies. Continued improvements in the cost and performance of wind power technologies have also been key drivers for wind capacity additions, yielding low-priced wind energy for utility, corporate, and other power purchasers.

This annual report—now in its fifteenth year—provides an overview of trends in the U.S. wind power market, with a particular focus on the year 2020. The report begins with an overview of installation-related trends: U.S. wind power capacity growth; how that growth compares to other countries and generation sources; the amount and percentage of wind energy in individual U.S. states; hybridization with storage and other sources of generation; and the quantity of proposed wind power capacity in interconnection queues in the United States. Next, the report covers an array of wind industry trends: developments in turbine manufacturer market share; manufacturing and supply-chain developments; wind turbine and component imports into the United States; project financing developments; and trends among wind power project owners and power purchasers. The report then turns to a summary of wind turbine technology trends: turbine capacity, hub height, rotor diameter, and specific power, as well as changes in site-average wind speed and recent repowering activity. After that, the report discusses wind performance, cost, and pricing. In doing so, it describes trends in capacity factors, wind turbine prices, installed project costs, and operations and maintenance (O&M) expenses. Levelized costs are calculated based on these input parameters. The report also reviews the prices paid for wind power through power purchase agreements (PPAs) and how those prices compare to the value of wind generation in wholesale energy markets, forecasts of future natural gas prices, and sales prices for . An additional comparison assesses the levelized cost of wind energy relative to its societal value, defined somewhat narrowly here to include the grid-system value of wind along with its health and climate benefits. Finally, the report concludes with a preview of possible near-term market developments based on the findings of other analysts.

Many of these trends vary by state or region, depending in part on the strength of the local wind resource. To that end, Figure 1 superimposes the boundaries of nine regions, seven of which align with organized wholesale power markets3, on a map of average annual U.S. wind speed at 80 meters above the ground. These nine regions will be referenced on many occasions throughout this report.

This edition of the annual report updates data presented in previous editions while highlighting recent trends and new developments. The report concentrates on larger, utility-scale wind turbines, defined here as individual turbines that exceed 100 kW in size.4 The U.S. wind power sector is multifaceted, and also includes smaller, customer-sited wind turbines used to power residences, farms, and businesses. Further information on distributed wind power, which includes smaller wind turbines as well as the use of larger turbines in distributed applications, is available through a separate annual report funded by the U.S. Department of Energy (DOE)—the Distributed Wind Market Report. In Chapters 2, 3, and 9—where it is sometimes difficult to separate offshore and land-based wind—this report emphasizes land-based and offshore wind, in combination. Other chapters exclusively focus on land-based wind. A companion study funded by DOE that focuses exclusively on is also available—the Offshore Wind Market Report.

3 The seven independent system operators (ISOs) include the Southwest Power Pool (SPP), Electric Reliability Council of Texas (ERCOT), Midcontinent Independent System Operator (MISO), California Independent System Operator (CAISO), ISO New England (ISO-NE), PJM Interconnection (PJM), and New York Independent System Operator (NYISO). 4 This 100-kW threshold between “smaller” and “larger” wind turbines is applied starting with 2011 projects to better match the American Clean Power Association’s historical methodology, and is also justified by the fact that the U.S. tax code makes a similar distinction. In years prior to 2011, different cut-offs are used to better match ACP’s reported capacity numbers and to ensure that older utility-scale wind power projects in California are not excluded from the sample.

1 Land-Based Wind Market Report

Sources: AWS Truepower, National Renewable Energy Laboratory (NREL) Figure 1. Regional boundaries overlaid on a map of average annual wind speed at 80 meters Much of the data included in this report were compiled by DOE’s Lawrence Berkeley National Laboratory (Berkeley Lab) from a variety of sources, including the U.S. Energy Information Administration (EIA), the Federal Energy Regulatory Commission (FERC), and the American Clean Power Association (ACP—along with its predecessor, the American Wind Energy Association). The Appendix provides a summary of the many data sources. In some cases, the data shown represent only a sample of actual wind power projects installed in the United States; furthermore, the data vary in quality. Emphasis should therefore be placed on overall trends, rather than on individual data points. Finally, each section of this report primarily focuses on historical and recent data. With some limited exceptions—including the final section of the report—the report does not seek to forecast wind energy trends.

2 Land-Based Wind Market Report

2 Installation Trends Wind power capacity grew at a record pace in 2020, with 16,836 MW of new capacity added in the United States and $24.6 billion invested

U.S. wind capacity additions equaled 16,836 MW in 2020, bringing the cumulative total to 121,985 MW at the end of the year (Figure 2).5 This growth represented $24.6 billion of investment in new wind power project installations in 2020, for a cumulative investment total of roughly $240 billion since the beginning of the 1980s.6,7 More than 70% of the new wind capacity installed in 2020 is located in ERCOT, MISO and SPP.

A relatively new trend is that of partial wind project repowering, in which major components of turbines are replaced. Such efforts provide access to favorable tax incentives, increase energy production with more- advanced turbine technology, and extend project life. In addition to the newly installed capacity reported above, 3,087 MW of existing wind plants were partially repowered in 2020, mostly in the form of increased rotor diameters and the replacement of major nacelle components.8

Annual Regional Capacity (GW) Cumulative Total Capacity (GW) 20 160 Noncontiguous Southeast (non-ISO) ISO-NE 15 NYISO 120 CAISO Total PJM West (non-ISO) MISO 10 80 SPP ERCOT

5 40

0 0 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 Source: ACP Figure 2. Annual and cumulative growth in U.S. wind power capacity As in previous years, growth was driven in part by continued improvements in the cost and performance of wind power technologies. The federal PTC, state renewables portfolio standards (RPS), and corporate demand

5 The 121,985 MW of cumulative capacity includes the 30 MW Block Island offshore wind ; the cumulative total for just land- based wind is 121,955 MW. When reporting annual capacity additions, this report focuses on gross additions, and does not consider partial repowering. The net increase in capacity each year can be somewhat lower, reflecting turbine decommissioning, or higher, reflecting partial repowering that increases nameplate capacities. Cumulative capacity (‘Total’ in Figure 2) includes both decommissioning and repowering. 6 All cost and price data are reported in real 2020 dollars. 7 These investment figures are based on an extrapolation of the average project-level capital costs reported later in this report and do not include investments in manufacturing facilities, research and development expenditures, or O&M costs; nor do they include investments to partially repowered plants. 8 The 3,087 MW of partially repowered capacity reflects the initial capacity, prior to refurbishment. Any change in capacity from partial repowering is included in the cumulative data but not the annual data reported in Figure 2.

3 Land-Based Wind Market Report for renewable energy also played important roles. Meanwhile, the ability of partially repowered wind projects to access the PTC has been the primary motivator for the growth in partial repowering in recent years.

Wind power represented the largest source of U.S. electric-generating capacity additions in 2020

Wind power has comprised a sizable share of generation capacity additions in recent years. In 2020, it constituted 42% of all U.S. capacity additions, moving ahead of solar power for the first time in several years (Figure 3).9 This occurred despite a record-breaking year for solar capacity additions, while natural gas capacity additions fell to a five-year low.

Annual Capacity Additions (GW)

40

20% 42% 30 26% 41% 24% 31%

20 29% 7% 24% 39%

10

0 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Wind Solar Other renewable Gas Coal Other non-renewable Sources: ABB, ACP, Wood Mackenzie, Berkeley Lab Figure 3. Relative contribution of generation types in annual capacity additions Over the last decade, wind power represented 29% of total U.S. capacity additions, and an even larger fraction of new generation capacity in SPP (75%), ERCOT (54%), MISO (52%), and the non-ISO West (32%) (Figure 4; see Figure 1 for regional definitions). Wind power’s contribution to generation capacity growth over the last decade is somewhat smaller—but still significant—in ISO-NE (13%), PJM (12%), NYISO (11%), and CAISO (10%), and considerably less in the Southeast (1%).

9 Data presented here are based on gross capacity additions, not considering retirements or partial repowering. Furthermore, they include only the 50 U.S. states, not U.S. territories.

4 Land-Based Wind Market Report

Percent of Capacity Additions: 2011−2020 100%

80%

60%

40%

20%

75% 54% 52% 32% 13% 12% 11% 10% 1% 29% 0% SPP ERCOT MISO West ISO-NE PJM NYISO CAISO Southeast U.S. (non-ISO) (non-ISO) Total* Wind Solar Other renewable Natural Gas Coal Other non-renewable *U.S. Total also includes AK and HI, in addition to the regions listed Sources: ABB, ACP, Wood Mackenzie, Berkeley Lab Figure 4. Generation capacity additions by region over last ten years

Globally, the United States ranked second in annual wind capacity additions in 2020, but remained well behind the market leaders in wind energy penetration

Global wind additions hit a new record in 2020, with nearly 93 GW of newly added capacity (including both land-based and offshore wind). With its 16.8 GW representing 18% of new global installed capacity in 2020, the United States continued to maintain its second-place position behind China (Table 1). Cumulative global installed wind capacity totaled 742 GW at the end of the year (GWEC 2021),10 with the United States accounting for 16%—a distant second to China. The United States also remains in second place, behind China, in annual wind electricity generation.

10 Yearly and cumulative installed wind power capacity in the United States are from the present report, while global wind power capacity comes from GWEC (2021) but are updated, where necessary, with the U.S. data presented here. Some disagreement exists among these data sources and others.

5 Land-Based Wind Market Report

Table 1. International Rankings of Wind Power Capacity: Land-Based and Offshore

Annual Capacity Cumulative Capacity (2020, MW) (end of 2020, MW) China 52,000 China 288,320 United States 16,836 United States 121,985 Brazil 2,297 Germany 62,850 Netherlands 1,979 India 38,625 Germany 1,668 Spain 27,250 Norway 1,532 United Kingdom 23,937 Spain 1,400 France 17,948 France 1,318 Brazil 17,750 Turkey 1,224 Canada 13,578 India 1,119 Italy 10,543 Rest of World 11,538 Rest of World 119,572 TOTAL 92,910 TOTAL 742,357 Sources: GWEC (2021); ACP for U.S.

A number of countries have achieved relatively high levels of wind energy penetration (i.e., wind generation as a percentage of total generation) in their electricity grids. Figure 5 presents data on a subset of countries. Wind penetration was nearly 50% in Denmark in 2020, and was between 25% and 40% in Ireland, Germany, the U.K, and Portugal. In the United States, wind supplied 8.3% of total electricity generation in 2020 (see Table 2 for additional details).

Wind as Percentage of Total Generation in 2020

50%

40%

30%

20%

10%

0% EU Italy U.K. India Brazil Spain China Turkey Ireland Austria France Poland Mexico Cyprus Croatia Greece Finland Estonia Canada Belgium Sweden Portugal Australia Romania Denmark Lithuania Germany Netherlands United States

Source: ACP Figure 5. Wind energy penetration in subset of top global wind markets

6 Land-Based Wind Market Report

Texas installed the most capacity in 2020 with 4,137 MW, while sixteen states exceeded 10% wind energy penetration as a fraction of total in-state generation

New utility-scale wind turbines were installed in 25 states in 2020. Texas once again installed the most new wind capacity of any state, adding 4,137 MW. As shown in Figure 6 and in Table 2, other leading states—in terms of new capacity—included Iowa, Oklahoma, Wyoming, Illinois, and Missouri, all of which added more than 1,000 MW (i.e., 1 GW) of new wind in 2020.

On a cumulative basis, Texas remained the clear leader, with 32,686 MW installed at the end of 2020—almost three times as much as the next-highest state (Iowa). In fact, Texas has more wind capacity than all but four countries (Table 1). States distantly following Texas in cumulative installed capacity include Iowa (>11 GW), Oklahoma (>9 GW), Kansas (>7 GW), Illinois (>6 GW), and California (almost 6 GW). Thirty-five states, plus Puerto Rico, had more than 100 MW of wind capacity as of the end of 2020, with 20 of these above 1 GW, 17 above 2 GW, and 11 above 3 GW.

Sources: ACP, Berkeley Lab Figure 6. Location of wind power development in the United States Some states have reached high levels of wind energy penetration. The right half of Table 2 lists the top 20 states based on actual wind electricity generation in 2020 divided by total in-state electricity generation and by in-state electricity sales in 2020. Electric transmission networks enable most states to both import and export power in real time, and states do so in varying amounts. Denominating in-state wind generation as both a proportion of in-state generation and as a proportion of in-state sales is relevant, but both should be viewed with some caution given varying amounts of imports and exports. As a fraction of in-state generation, Iowa leads the list, with 57.3% of electricity generated in the state coming from wind, followed by Kansas, Oklahoma, South Dakota, and North Dakota. As a fraction of in-state sales, Iowa is also the leading state, with

7 Land-Based Wind Market Report

68.6% of the electricity sold in the state being met by wind, followed by North Dakota and Kansas (both also over 60%), and then Oklahoma and South Dakota (both over 40%). Sixteen states have achieved wind penetration levels of 10% or higher when expressed either as a percentage of generation or as a percentage of sales.

Table 2. U.S. Wind Power Rankings: The Top 20 States

Installed Capacity (MW) 2020 Wind Generation as a Percentage of: Annual (2020) Cumulative (end of 2020) In-State Generation In-State Sales Texas 4,137 Texas 32,686 Iowa 57.3% Iowa 68.6% Iowa 1,215 Iowa 11,377 Kansas 43.2% North Dakota 61.7% Oklahoma 1,162 Oklahoma 9,344 Oklahoma 35.4% Kansas 61.2% Wyoming 1,072 Kansas 7,028 South Dakota 32.9% Oklahoma 47.9% Illinois 1,059 Illinois 6,409 North Dakota 30.8% South Dakota 44.8% Missouri 1,028 California 5,922 Maine 23.8% Wyoming 33.5% Colorado 967 Colorado 4,692 Nebraska 23.6% Nebraska 28.8% Kansas 896 Minnesota 4,291 Colorado 23.2% New Mexico 28.6% South Dakota 774 North Dakota 3,989 Minnesota 21.6% Texas 22.7% New Mexico 770 Oregon 3,737 New Mexico 20.7% Colorado 22.6% Indiana 651 Washington 3,395 Texas 19.5% Maine 22.3% Michigan 493 Indiana 2,968 Vermont 15.1% Montana 20.8% Minnesota 455 New Mexico 2,723 Idaho 14.1% Minnesota 19.6% Nebraska 393 Wyoming 2,687 Oregon 13.1% Oregon 18.0% Arizona 350 Michigan 2,681 Montana 12.6% Illinois 13.0% North Dakota 343 Nebraska 2,531 Wyoming 12.3% Idaho 11.3% Oregon 310 South Dakota 2,305 Illinois 9.8% Washington 9.8% Washington 287 Missouri 1,987 Washington 7.3% Vermont 7.3% Ohio 126 New York 1,987 Indiana 7.3% Michigan 7.0% California 111 Pennsylvania 1,459 California 6.4% Indiana 7.0% Rest of U.S. 238 Rest of U.S. 7,786 Rest of U.S. 1.2% Rest of U.S. 1.0% TOTAL 16,836 TOTAL 121,985 TOTAL 8.3% TOTAL 9.2% Note: Based on 2020 wind and total generation and retail sales by state from EIA’s Electric Power Monthly. Sources: ACP, EIA Given the ability to trade power across state boundaries, estimates of wind penetration within entire multi-state markets operated by the major independent system operators (ISOs) are also relevant. In 2020, wind penetration (expressed as a percentage of load) was 31.3% in SPP, 22.7% in ERCOT, 11.0% in MISO, 6.6% in CAISO, 3.4% in PJM, 3.0% in ISO-NE, and 2.9% in NYISO. A small but growing number of hybrid plants that pair wind with storage and other resources are operating in the United States

There were 38 hybrid wind power plants in operation at the end of 2020, representing 2.3 GW of wind and 0.9 GW of co-located assets (storage, PV, or fossil-fueled generators). Some of these represent full hybrids where, for example, wind and storage are co-located and the design, configuration, and operation of the constituent technologies are fully integrated. In other cases, plants are co-located, sharing a point of interconnection, but are designed, configured, and operated more independently (e.g., hybrids that pair wind and gas plants).

The most common type of wind hybrid project combines wind and storage technology, where 1.4 GW of wind has been paired with 0.2 GW of battery storage across 14 plants. Other combinations include wind and PV; wind, PV, and storage; wind and gas; and more (Figure 7). The ERCOT region hosts the largest amount of wind capacity in hybrid plants (0.86 GW), followed by PJM (0.66 GW) and the non-ISO West (0.42 GW). Wind capacity tends to be larger for wind+storage hybrids than for other hybrid configurations.

8 Land-Based Wind Market Report

Sources: EIA-860 2020 Early Release, Berkeley Lab

Figure 7. Location and capacity of hybrid wind plants in the United States Figure 8 displays design characteristics for a subset of the more-common hybrid plant configurations, including those that do not incorporate wind. Wind+storage hybrids have a 14% storage-to-generator ratio with an average storage duration of just 0.6 hours, suggesting a focus on providing ancillary services and only limited capacity to shift large amounts of energy across time. Fossil+storage hybrids have the smallest storage- to-generator ratio (7%) but longer battery durations (1.3 hours). PV+storage hybrids have higher average storage-to-generator ratios (25%) and battery durations (2.6 hours). Based on data from proposed projects, presented in the next section on interconnection queues, there is growing interest in hybridizing with larger storage-to-generator ratios and longer storage durations.

# projects Total capacity (MW) Storage ratio Duration (hrs)

0 500Wind 1,000PV 1,500 Fossil2,000 Storage2,500 3,000 3,500 4,000 4,500

PV+Storage 73 991.6 249.7 Wind 25% 2.6 PV Fossil Wind+Storage 14 1,425.3 197.6 14% 0.6 Storage

Wind+PV+Storage 2 217.5 20.5 34.0 14% 0.4

Fossil+Storage 21 4,003.1 281.2 7% 1.3

Wind+PV 7 585.7 266.5 0.0 n/a n/a

Notes: Not included in the figure are 111 hybrid projects with other configurations. Storage ratio defined as storage capacity divided by total generator capacity. Sources: EIA-860 2020 Early Release, Berkeley Lab Figure 8. Design characteristics of hybrid power plants operating in the United States, for a subset of configurations

9 Land-Based Wind Market Report

The trend to co-locate wind with other assets has progressed at a steady pace since 2006, but no new wind hybrids commenced operation in 2020 whereas 28 new PV+storage projects came online. Despite a slight contraction since 2018, substantial wind power capacity exists in transmission interconnection queues; solar and storage reached new highs in 2020

One testament to the amount of developer and purchaser interest in wind energy is the amount of wind power capacity working its way through the major transmission interconnection queues across the country. Figure 9 provides this information over the last six years for wind power and other resources aggregated across 42 different interconnection queues administered by ISOs and utilities.11 These data should be interpreted with caution: placing a project in the interconnection queue is a necessary step in project development, but being in the queue does not guarantee that a project will be built. An analysis of five ISO queues found an overall average completion rate of 24% for projects of all types proposed from 2000 to 2015 (Rand et al. 2021). Some projects are speculative in , and duplicate projects also complicate interpretation.

Capacity in Queues at Year-End (GW) 500 Entered queues in the year shown Entered queues in an earlier year

400 Hatched portion indicates the amount paired with storage

300 For storage, hatched portion indicates the amount paired with generation 200

100

0 2015 - 2020 2015 - 2020 2015 - 2020 2015 - 2020 2015 - 2020 2015 - 2020 2015 - 2020 Storage Gas Nuclear Coal Other Notes: Data on hybrid projects paired with storage were not collected for years prior to 2018. Additionally, the ‘storage’ data for 2018 reflects stand-alone storage (not hybrid plants), whereas for years prior to 2018 the storage data also include some hybrid facilities. Hybrid storage capacity is estimated using storage:generator ratios from projects that provide separate capacity data. Storage capacity in hybrids was not estimated for years prior to 2020. Source: Berkeley Lab review of interconnection queues Figure 9. Generation capacity in 42 selected interconnection queues from 2015 to 2020, by resource type Even with this important caveat, the amount of wind capacity in the nation’s interconnection queues still provides at least some indication of the amount of developer interest. At the end of 2020, there were 209 GW of wind power capacity in the interconnection queues reviewed for this report—a small decrease from the 225 GW in the queues the previous year and the peak (232 GW) at the end of 2018. In 2020, 55 GW of new wind capacity entered the queues, 10 GW of which were in hybrid configurations and 25 GW of which were for offshore wind. Solar additions to interconnection queues far outpaced wind in 2020, with 171 GW added.

11 The queues surveyed include PJM, MISO, NYISO, ISO-NE, CAISO, ERCOT, SPP, Western Area Power Administration (WAPA), Bonneville Power Administration (BPA), Tennessee Valley Authority (TVA), and a large number of other individual utilities. To provide a sense of sample size and coverage, the ISOs, RTOs, and utilities whose queues are included here have an aggregated non-coincident (balancing authority) peak demand of over 85% of the U.S. total. The figures in this section only include projects that were active in the queues at the times specified but that had not yet been built; suspended projects are not included.

10 Land-Based Wind Market Report

Storage additions to the queues have increased rapidly in recent years as well, both for standalone plants and hybridized with solar or wind. Overall, wind represented 28% of all generator capacity in the queues at the end of 2020, compared to 61% for solar and 10% for natural gas.

The wind capacity in the interconnection queues is spread across the United States, as shown in Figure 10, with the largest amounts in SPP (22%), NYISO (16%), the West (non-ISO) (16%), and PJM (14%). Smaller amounts are found in MISO (9%), ERCOT (8%), ISO-NE (7%), CAISO (6%), and the Southeast (non-ISO) (1%). A majority (57%) of wind capacity in the queues has requested to come online by the end of 2023, and 17% of wind capacity has a fully executed interconnection agreement.

NYISO, CAISO, and the West (non-ISO) experienced especially large annual additions in 2020, with NYISO’s additions being entirely for offshore wind. Across all queues, 29% (61 GW) of all wind capacity in the queues at the end of 2020 was offshore wind, and that share is growing: 45% (25 GW) of the wind added to queues in 2020 was offshore. Offshore wind capacity was added on both the East Coast (NYISO, PJM, ISO- NE) and the West Coast (CAISO).

Source: Berkeley Lab review of interconnection queues Figure 10. Wind power capacity in 42 selected interconnection queues, by region As shown in Figure 11, 34% of the solar capacity in interconnection queues at the end of 2020 has been proposed as hybrid plants, whereas only 6% of the wind capacity is paired with storage or another generation resource. Of the 13 GW of proposed wind capacity in hybrid configurations, the majority (7 GW) is paired with storage, with somewhat less paired with solar (4 GW) or both solar and storage (2 GW).

Capacity in Queues at end of 2020 (GW) Wind Hybrids Solar Hybrids 13 GW-wind total 159 GW-solar total 500 Standalone Solar+ Hybrid Wind 400 Solar+ Solar+ Nat. Gas 300 Wind+ Solar+ Wind+ Storage Solar+ Storage Hydro 200 Wind+ Wind+ Solar 34% Storage 100 55%

Solar+ 6% 6% 15% 0 Storage Solar Wind Storage Natural Gas Other

Source: Berkeley Lab review of interconnection queues

Figure 11. Hybrid power plants in 42 selected interconnection queues, including wind and solar hybrid configurations

11 Land-Based Wind Market Report

As shown in Figure 12, commercial interest in wind hybrid plants is highest in California and the West (non- ISO). In fact, 37% of the wind in CAISO’s queues is proposed as a hybrid, as is 13% of the wind in the West. Finally, focusing on the wind+storage hybrids, the storage:generator ratios are reported for a subset of the queues—CAISO, ERCOT, and SPP.12 These show a storage:generator ratio of 45% on average across these queues, ranging from an average of 28% in SPP to 99% in CAISO. These are lower ratios than for solar+storage hybrids, which average 66% across these queues.

Source: Berkeley Lab review of interconnection queues

Figure 12. Hybrid wind power plants in 42 selected interconnection queues at the end of 2020 As additional measures of the near-term development pipeline, ACP (2021b) reports that 34 GW of wind power capacity was under construction or at an advanced stage of development at the end of the first quarter of 2021. EIA (2021b) identifies 19 GW of planned additions for 2021 and 2022 combined.

12 Only regions that reported wind and storage capacities for three or more projects in the queues are shown.

12 Land-Based Wind Market Report

3 Industry Trends GE and Vestas supplied turbines for 87% of U.S. wind power capacity installed in 2020

Of the 16,836 MW of wind installed in the United States in 2020, GE Wind supplied 53%, with Vestas coming in second (34%), followed by Siemens Gamesa Renewable Energy (SGRE, 9%), Nordex (3%), and Goldwind (1%) (Figure 13).13 GE and Vestas have dominated the U.S. market for some time, with SGRE and—more recently—Nordex vying for third.

U.S. Market Share by MW 100% Other Goldwind 80% Acciona (pre-2016) Nordex (pre-2016) 60% Nordex Acciona Gamesa (pre-2017) 40% Siemens (pre-2017) SGRE 20% Vestas GE Wind 0% 2005 2008 2011 2014 2017 2020 Source: ACP Figure 13. Annual U.S. market share of wind turbine manufacturers by MW, 2005–2020

The domestic wind industry supply chain was reasonably stable in 2020

Despite COVID-19, with record growth in U.S. wind installations, wind-related job totals in the United States increased by 1.8% in 2020, to 116,800 full-time workers (DOE 2021). These jobs include, among others, those in construction (~42,300) and manufacturing (~23,900).

Figure 14 identifies the 535 wind turbine component manufacturing, assembly, and other supply chain facilities operating in the United States at the end of 2020. Three of the major turbine OEMs that serve the U.S. wind industry—GE, Vestas, and SGRE—are represented within this total, each having one or more operating manufacturing facility. The figure also highlights the geographic breadth of the domestic supply chain.

13 Market share is reported in MW terms and is based on project installations in the year in question.

13 Land-Based Wind Market Report

Source: ACP Figure 14. Location of turbine and component manufacturing facilities In aggregate, domestic turbine nacelle assembly14 capability—defined here as the maximum annual nacelle assembly capability of U.S. plants if all were operating at full utilization—grew from less than 1.5 GW in 2006 to more than 13 GW in 2012, fell to roughly 10 GW in 2015, and then rose to and has held steady at 15 GW (Figure 15). In addition, at the end of 2020, U.S. manufacturing facilities had the capability to produce 10,200 individual blades (~9.4 GW if using average sized turbines) and 3,800 towers (~10.3 GW) annually. Figure 15 contrasts this equipment manufacturing capability with past U.S. wind additions as well as near-term forecasts of future new installations (see Chapter 9, “Future Outlook”). It demonstrates that domestic manufacturing capability for blades, towers, and nacelle assembly has been reasonably well balanced against historical market demand, though record wind additions in 2020 ensured that demand for blades and towers outstripped domestic manufacturing capability last year. Manufacturing facilities do not typically operate at maximum capability; see the next section of the report for estimates of domestic manufacturing content.

14 Nacelle assembly is defined as the process of combining the multitude of components included in a turbine nacelle, such as the gearbox and generator, to produce a complete turbine nacelle unit.

14 Land-Based Wind Market Report

Capacity (GW) 15 Nacelle assembly Averages of deployment projections

10 Blades capacity Tower capacity

5

Annual installed wind power capacity 0 2006 2008 2010 2012 2014 2016 2018 2020 2022e 2024e

Sources: ACP, independent analyst projections, Berkeley Lab Figure 15. Domestic wind manufacturing capability vs. U.S. wind power capacity installations Finally, fierce competition throughout the supply chain has reduced turbine OEM profitability over the last several years (Figure 16).15

Profit Margin (EBITDA)

20% Goldwind Gamesa/SGRE 15%

Vestas 10% Nordex 5%

0%

-5%

-10% 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

Note: EBITDA = Earnings Before Interest, Taxes, Depreciation and Amortization; Sources: OEM annual reports and financial statements Figure 16. Turbine OEM global profitability

15 Although it is one of the largest turbine suppliers in the U.S. market, GE is not included because it is a multi-national conglomerate that does not report segmented financial data for its wind turbine division.

15 Land-Based Wind Market Report

Domestic manufacturing content is strong for some wind turbine components, but the U.S. wind industry remains reliant on imports

Despite the breadth of the domestic wind industry supply chain, the U.S. wind sector is reliant on imports of wind equipment. The level of dependence varies by component: some components have a relatively high domestic share, whereas others remain largely imported. These trends are revealed, in part, by data on wind equipment trade from the U.S. Department of Commerce.16

Figure 17 presents data on the dollar value of estimated imports to the United States of wind-related equipment that can be tracked through trade codes. The figure shows imports of wind-powered generating sets and parts, including nacelles (i.e., nacelles with blades, nacelles without blades, and, in some cases, other turbine components internal to the nacelle) as well as imports of other select turbine components shipped separately from the generating sets and nacelles.17 The turbine components included in the figure consist only of those that can be tracked through trade codes: towers, generators (as well as generator parts), and blades and hubs.18

Imports (Billion 2020$) Other wind-related equipment Wind generators and generator parts 6 Wind blades and hubs Wind towers Wind-powered generating sets and parts, including nacelles

4

2

0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Note: Wind-related trade codes and definitions are not consistent over the full time period. Source: Berkeley Lab analysis of data from USA Trade Online, https://usatrade.census.gov Figure 17. Imports of wind-related equipment that can tracked with trade codes

16 See the Appendix for further details on data sources and methods used in this section, including the specific trade codes considered. 17 Wind turbine components such as blades, towers, and generators are included in the data on wind-powered generating sets and nacelles if shipped in the same transaction. Otherwise, these component imports are reported separately. 18 Though all of the import estimates are specific to wind equipment in 2020, import trends should be viewed with particular caution because the underlying data from earlier years used to produce Figure 17 are based on trade categories that are not all exclusive to wind. Some of the import estimates shown in Figure 17 for years prior to 2020 therefore required assumptions about the fraction of larger trade categories likely to be represented by wind turbine components. For example, from 2013 through 2019, nacelles (when shipped alone) are included in a trade category that is not largely exclusive to wind. The trade code for tower imports is also not entirely exclusive to wind, but is believed to be dominated by wind since 2011—we assume that 100% of imports from this trade category, since 2011, represent wind equipment.

16 Land-Based Wind Market Report

The estimated imports of tracked wind-related equipment into the United States increased substantially from 2006 to 2008, before falling through 2010, increasing somewhat in 2011 and 2012, and then plummeting in 2013 with the simultaneous drop in U.S. wind installations. From 2014 through 2020, imports of wind-related turbine equipment generally followed U.S. wind installation trends, bouncing back from the low of 2013. Because imports of component parts occur in additional, broad trade categories different from those included in Figure 17, the data presented here understate the aggregate amount of wind equipment imports.

Figure 18 shows the total value of tracked wind-specific imports to the United States in 2020, by country of origin, as well as states of entry. Major countries from which the U.S. imports wind equipment include India, Spain, China, and Mexico, each with over $500 million in wind-specific exports to the U.S. in 2020. Texas is the dominant entry point, with almost $3 billion of wind-specific equipment flowing through it in 2020, followed distantly by Florida, Michigan, and Minnesota.

Note: Line widths are proportional to import amount by country. Figure does not intend to depict the destination of these imports, by state. Source: Berkeley Lab analysis of data from USA Trade Online, https://usatrade.census.gov

Figure 18. Summary map of tracked wind-specific imports in 2020: top-10 countries of origin and states of entry Looking behind these data, India, followed by Denmark, Germany, Brazil, and Spain, were the primary source countries for wind-powered generating sets and parts, including nacelles, in 2020 (Figure 19). Tower imports came from a mix of countries near and far—Canada, India, Malaysia, Spain, and Indonesia. With regard to blades and hubs, China, Mexico, India, Spain and Brazil were the largest source countries in 2020. Finally, wind-related generators and generator parts primarily came from Spain, Vietnam, Germany and Serbia.

17 Land-Based Wind Market Report

Wind-powered generating sets and parts, including nacelles Wind generators and parts

Total 2020 imports: Total 2020 imports: $1,099 million $211 million

Top countries: Top countries: India (26%) Spain (35%) Denmark (15%) Vietnam (30%) Germany (15%) Germany (13%) Asia Brazil (10%) Serbia (11%) Spain (8%) Denmark (1%) Europe

North America Wind towers Wind blades and hubs South America

Other Total 2020 imports: Total 2020 imports: $730 million $2,611 million

Top countries: Top countries: Canada(20%) China (19%) India (19%) Mexico(18%) Malaysia (16%) India (14%) Spain (14%) Spain (13%) Indonesia (10%) Brazil (10%)

Source: Berkeley Lab analysis of data from USA Trade Online, https://usatrade.census.gov Figure 19. Origins of U.S. imports of selected wind turbine equipment in 2020

Figure 20 presents rough estimates of the domestic content for a subset of the major wind turbine components used in new (and repowered) U.S. wind projects in 2020. Domestic content is relatively strong for larger components such as towers and blades, though domestic content in blades has declined in recent years. Nacelle assembly also has high domestic content, wherein domestic and imported components are assembled into complete nacelles on U.S. .

These figures understate the wind industry’s reliance on foreign suppliers, because significant wind-related imports occur under trade categories not captured in this figure, including equipment (such as mainframes, converters, pitch and yaw systems, main shafts, bearings, bolts, controls) and manufacturing inputs (such as foreign steel in domestic manufacturing). BloombergNEF (2021b) has recently estimated that a typical onshore wind project in the U.S. sources 57% of its components (by dollar value) domestically.

18 Land-Based Wind Market Report

Nacelle Assembly >85%

Wind Towers 60−75%

Blades and Hubs 30−50%

0% 20% 40% 60% 80% 100% Domestic Content Source: Berkeley Lab analysis

Figure 20. Approximate domestic content of major components in 2020

Project finance was volatile in 2020

The renewable energy project finance market was volatile in 2020, due in large part to the effects of the pandemic. In response to the initial wave of lockdowns imposed in mid-March, the Federal Reserve slashed the overnight Federal Funds Rate by 100 basis points, and the London Interbank Offered Rate (LIBOR) soon followed suit, heading down to around 0%. Swap rates also initially moved a bit lower. Bank margins, on the other hand, increased by around 50 basis points in response to the unfolding uncertainty (Norton Rose Fulbright 2021). By late summer, however, bank spreads had mostly returned to pre-pandemic levels, and swap rates started to move upwards, dragging the all-in cost of term debt higher. By the end of the year, all-in term debt was priced only slightly lower than where it had started the year (Figure 21).

Tax equity investors continued to honor their many commitments in what was to have been the last year in which projects placed in service would qualify for 100% of the PTC. The result was a record high $11+ billion of third-party tax equity invested in new and repowered wind projects in 2020—several billion dollars more than in prior years (Norton Rose Fulbright 2021). Nonetheless, new tax equity commitments reportedly grew more scarce as the year dragged on, eventually leading to higher yields of around 7% by the end of the year— roughly 50 basis points higher than at the start of the year. The tax equity shortage has continued into 2021, bolstering calls for the return of a “direct pay” alternative, perhaps in the form or a refundable PTC or along the lines of the “Section 1603” cash grant that was available in lieu of the PTC (or ITC) from 2009–2012.

19 Land-Based Wind Market Report

12%

10%

Tax equity yield (after-tax) 8%

6% 15-year debt interest rate (pre-tax) 4%

2% 15-year debt interest rate (after-tax) 0% 2005 2007 2009 2011 2013 2015 2017 2019 2021 Sources: Intercontinental Exchange Benchmark Administration, BNEF (2017), Norton Rose Fulbright (2021) Figure 21. Cost of 15-year debt and tax equity for wind projects over time In the meantime, the PTC remains in force, and has recently been extended on several occasions and in several ways. In December 2019, Congress extended the PTC’s construction start deadline through 2020 and restored its level to 60% of the full credit (it had been at 40% for projects starting construction in 2019). In May 2020, pandemic-related supply chain disruptions led the IRS to tack an additional year onto the original 4-year safe harbor window for projects that started construction in 2016 or 2017. This effectively meant that projects that started construction in 2016 had until the end of 2021 to be placed in service and receive 100% of the PTC, while projects that started construction in 2017 had until the end of 2022 to claim 80% of the PTC. In December 2020, Congress extended the PTC for yet another year (i.e., projects must start construction by the end of 2021) at the 60% level, while in June 2021, the IRS extended the safe-harbor window yet again, to six years for projects that started construction in 2016–2019 and to five years for projects that started construction in 2020. With these newly extended safe harbor windows, projects that started construction in 2016 can now claim 100% of the PTC as long as they achieve commercial operations by the end of 2022; this drops to 80% of the PTC for projects that started construction in 2017 and come online by the end of 2023, and then 60% of the PTC for projects that started construction in 2019–2021 and achieve commercial operations by the end of 2025. Finally, there are active discussions among lawmakers surrounding possible additional PTC extensions—which could have a significant impact on the project finance market in years to come.

Independent power producers own the majority of wind assets built in 2020

Independent power producers (IPPs) own 12,344 MW or 73% of the 16,836 MW of new wind capacity installed in the United States in 2020 (Figure 22, right pie chart). Investor-owned utilities (IOUs)—most notably Xcel Energy (1,420 MW), Alliant/Interstate Power & Light (532 MW), MidAmerican (486 MW) and Pacificorp (453 MW), but including thirteen IOUs in all—installed a total of 4,492 MW (27%). Of the cumulative installed wind power capacity at the end of 2020 (Figure 22, left chart), IPPs own 81% and utilities

20 Land-Based Wind Market Report own 17% (16% IOU and 1% publicly owned utility, or POU), with the remaining 2% falling into the “other” category of projects owned by neither IPPs nor utilities (e.g., owned by towns, schools, businesses, farmers).19

% of Cumulative Installed Capacity Other 2020 Capacity by 100% Owner Type Investor-Owned Utility (IOU)

80% Publicly Owned Utility (POU) IPP: 12,344 MW 60%

40% IOU: Independent Power Producer (IPP) 4,492 MW 20%

0% 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020

Source: Berkeley Lab estimates based on ACP Figure 22. Cumulative and 2020 wind power capacity categorized by owner type

Long-term contracted sales to utilities remained the most common offtake arrangement, but direct retail sales and merchant offtake arrangements were both significant

Electric utilities continued to be the largest offtakers of wind power (i.e., ‘users’ of wind to serve load) in 2020 (Figure 23, right pie chart), either owning wind projects (27%) or buying the electricity from wind projects (32%) that, in total, represent 59% of the new capacity installed last year (with the 59% split between 45% IOU and 14% POU). On a cumulative basis, utilities own (18%) or buy (44%) power from 62% of all wind power capacity installed in the United States (with the 62% split between 43% IOU and 19% POU, with the POU category including community choice aggregators (CCAs)).

19 Many of the “other” projects, along with some IPP- and POU-owned projects, might also be considered “community wind” projects that are owned by or benefit one or more members of the local community to a greater extent than typically occurs with a commercial wind project.

21 Land-Based Wind Market Report

% of Cumulative Installed Capacity Direct Retail On-Site 100% 2020 Capacity by Offtake Type

Power Marketer Merchant/Quasi-Merchant 80% Merchant: IOU: 2,007 MW 7,521 MW 60% POU Retail: 40% 4,496 MW POU: 2,406 MW

20% IOU Power Marketer: 406 MW 0% Undisclosed 252 MW 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 (3%) Source: Berkeley Lab estimates based on ACP Figure 23. Cumulative and 2020 wind power capacity categorized by power offtake arrangement Direct retail purchasers of wind power, including a diverse and growing set of corporate and non-corporate offtakers, are supporting 4,496 MW or 27% of the new wind power capacity installed in the United States in 2020 (and 11% of cumulative wind power capacity). Merchant/quasi-merchant projects accounted for 12% of all new 2020 capacity and 21% of cumulative capacity.20 Finally, power marketers—defined here to include commercial intermediaries that purchase power under contract and then resell that power to others21—are buying the remaining 2% of new 2020 wind capacity and 5% of cumulative capacity.

20 Merchant/quasi-merchant projects are those whose electricity sales revenue is tied to short-term contracts and/or wholesale spot electricity market prices (with the resulting price risk commonly hedged over a 10- to 12-year period), rather than being locked in through a long-term PPA. Most of these projects are located within ERCOT in Texas, though there are some merchant/quasi-merchant projects within other markets, including PJM, MISO, SPP, and NYISO. Associated hedges are often structured as a “fixed-for-floating” power price swap—a purely financial arrangement whereby the wind power project swaps the “floating” revenue stream that it earns from spot power sales for a “fixed” revenue stream based on an agreed-upon strike price with the swap counterparty. 21 These intermediaries include the wholesale marketing affiliates of large IOUs, which may buy wind on behalf of their load-serving affiliates.

22 Land-Based Wind Market Report

4 Technology Trends Turbine capacity, rotor diameter, and hub height have all increased significantly over the long term

The average nameplate capacity of newly installed wind turbines in the United States in 2020 was 2.75 MW, 8% larger than in 2019 and up 284% since 1998–1999 (Figure 24).22 The average hub height of turbines installed in 2020 was 90.1 meters, unchanged from 2019 but up 59% since 1998–1999. The average rotor diameter in 2020 was 124.8 meters, 3% larger than in 2019 and up 159% since 1998–1999 (translating to a 570% growth in rotor swept area since 1998-1999). Trends in rotor scaling in particular, but also hub height, are two of several factors impacting the project-level capacity factors highlighted later in this report.

Capacity (MW) Height & Diameter (m) 3.5 140 Rotor diameter 3.0 120

2.5 100 Hub height 2.0 80

1.5 60

1.0 40

0.5 20 Nameplate capacity 0.0 0 '98−99 '02−03 2006 2008 2010 2012 2014 2016 2018 2020 Sources: ACP, Berkeley Lab Figure 24. Average turbine nameplate capacity, hub height, and rotor diameter for land-based wind projects Figure 25 presents these same trends since 2010, but with additional detail on the relative distribution of turbines with different capacities, hub heights, and rotor diameters. For example, 2020 saw a sharp increase in the proportion of turbines installed in the 2.75–3.5 MW range, while the proportion of turbines at 3.5 MW or larger also increased. The percentage of turbines with hub heights larger than 100 meters increased in 2020, to 15%, but the share of turbines with 90–100 meter hub heights declined somewhat, as 80–90 meter hub heights were better-represented. Finally, the steady progression toward larger rotors has dominated the industry for some time. In 2010, no turbines employed rotors that were 115 meters in diameter or larger, while 91% of newly installed turbines featured such rotors in 2020 (and 17% of those were at least 130 meters).

22 Figure 24 and a number of the other figures and tables included in this report combine data into both one- and two-year periods in order to avoid distortions related to small sample size in the PTC lapse years of 2000, 2002, and 2004; although not a PTC lapse year, 1998 is grouped with 1999 due to the small sample of 1998 projects. Though 2013 was a slow year for wind additions, it is shown separately here despite the small sample size.

23 Land-Based Wind Market Report

% of turbines Average Capacity (MW)% of Turbines Average Hub height (m)% of Turbines Average Rotor diameter (m) 100% 3.0100% 95100% 140

80% 2.580% 9080% 120 average average 60% average 2.060% 8560% 100

40% 1.540% 8040% 80 ≥3.5 MW ≥100 m ≥130 m 2.75−3.5 MW 90−100 m 115-130 m 20% 75 20% 2.0−2.75 MW 1.0 80−90 m 20% 100-115 m 60 <2.0 MW <80 m <100 m 0% 0.5 0% 700% 40 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Sources: ACP, Berkeley Lab Figure 25. Trends in turbine nameplate capacity, hub height, and rotor diameter

Turbines originally designed for lower wind speed sites dominate the market

The growth in the average swept area (in m2) of rotors has been especially rapid over the last two decades, outpacing growth in average nameplate capacity (in W). This has resulted in a decline in the average “specific power” (in W/m2) among the U.S. turbine fleet over time, from 393 W/m2 among projects installed in 1998– 1999 to 223 W/m2 among projects installed in 2020 (Figure 26).

All else equal, a lower specific power will boost capacity factors, because there is more swept rotor area available (resulting in greater energy capture) for each watt of rated turbine capacity. This means that the generator is likely to run closer to or at its rated capacity more often. In general, turbines with low specific power were originally designed for lower wind speed sites, intended to maximize energy capture in areas where large-rotor machines would not be placed under excessive physical stress due to high or turbulent winds. As suggested in Figure 26 and as detailed later, however, such turbines are in widespread use in the United States—even in sites with relatively high wind speeds. The impact of lower specific-power turbines on project- level capacity factors is discussed in more detail in Chapter 5.

% of Turbines Average Specific power (W/m2)

100% 400

80% 350

average 60% 300 ≥350 ≥300−350 40% ≥250−300 250 ≥200−250 180−200 20% 200

0% 150 '98−99 '02−03 2006 2008 2010 2012 2014 2016 2018 2020

Sources: ACP, Berkeley Lab Figure 26. Trends in turbine specific power

24 Land-Based Wind Market Report

Wind turbines were deployed in somewhat lower wind-speed sites in 2020 than in the previous seven years

Figure 27 shows the long-term average wind resource at wind project sites, by commercial operation date. The figure depicts both the long-term site-average wind speed (in meters per second) at 80 meters for projects installed in each year (right scale) and an index of wind resource quality at 80 meters (left scale).23

Wind plants that came online in 2020 are located—on average—at sites with an estimated long-term average 80-meter wind speed of 7.7 meters per second (m/s). This is the lowest average site wind speed in the last eight years. Federal Aviation Administration (FAA) and industry data on projects that are either “under construction” or in “advanced development” suggest that the sites likely to be built out over the next few years will, on average, have even lower long-term average wind speeds. Projects potentially built after that, characterized as either “pending” or “proposed,” appear destined for sites with slightly higher or similar wind speeds.24 Trends in the wind resource quality index—which represents estimates of the gross capacity factor for each turbine location, indexed to the 1998–1999 installations—are similar.

Several factors could have driven these observed trends in average site quality. First, the increased availability of low-wind-speed turbines that feature lower specific power have enabled the economic build-out of lower- wind-speed sites. Second, transmission constraints (or other siting constraints, or even just regionally differentiated wholesale electricity prices) may have, over time, increasingly focused developer attention on those projects in their pipeline that have access to transmission (or higher-priced markets, or readily available sites without long permitting times), even if located in somewhat lower wind resource sites. The build-out of new transmission (for example, the completion of major transmission additions in West Texas in 2013), however, may at times have offered the chance to install new projects in more energetic sites. Other forms of federal and/or state policy could also play a role. For example, wind projects built in the four-year period from 2009 through 2012 were able to access a 30% cash grant (or ITC) in lieu of the PTC. Many projects availed themselves of this incentive and, because the dollar amount of the grant (or ITC) was not dependent on how much electricity a project generates, it is possible that developers also seized this limited opportunity to build out the less-energetic sites in their development pipelines. Finally, state policies sometimes motivate in-state or in-region wind development in lower wind resource regimes.

Increasing hub heights help to partially offset these trends in 80-meter wind speeds and resource quality. From 2014 to 2020, roughly half of the decline in 80-meter wind speeds has been offset by increasing hub heights.

23 The wind resource quality index is based on site estimates of gross capacity factor at 80 meters by AWS Truepower. A single, common wind turbine power curve is used across all sites and timeframes in this case, and no losses are assumed. The values are indexed to projects built in 1998—1999. Further details are found in the Appendix. A benefit of this wind resource quality index is that changes in the index value will better approximate expected changes in actual wind project performance than will changes in average annual wind speed. 24 “Under construction” turbines are part of a project where construction has begun, but the project has not yet been commissioned. Turbines in “advanced development” have one of the following in place: a signed PPA (or similar long-term contract), a firm turbine order, or an announcement to proceed under utility ownership, indicating a high-likelihood that they will be built. “Pending” turbines are those that have received a “No Hazard” determination by the FAA and are not set to expire for another 18 months, while “proposed” turbines have not yet received any determination. Pending and proposed turbines may not all ultimately be built. However, analysis of past data suggests that FAA pending and proposed turbines offer a reasonable proxy for turbines built in subsequent years.

25 Land-Based Wind Market Report

Wind resource quality (index, 80m) Wind speed (m/s, 80m) 100 8.6

95 8.4

Past Future 90 8.2

85 8.0

80 7.8

Wind 75 7.6 Speed

70 7.4 Index of wind resource quality Under Const. Adv. Dev. Pending 65 Proposed 7.2 98-99 02-03 2006 2008 2010 2012 2014 2016 2018 2020 Sources: ACP, Berkeley Lab, AWS Truepower, FAA Obstacle Evaluation / Airport Airspace Analysis files Figure 27. Wind resource quality by year of installation at 80 meters

Low-specific-power turbines are deployed on a widespread basis; taller towers are seeing increased use in a wider variety of sites

One might expect that the increasing market share of low-specific-power turbines would be due to a movement by developers to deploy turbines in lower wind speed sites. There is some evidence of this movement historically (see Figure 27), but it is clear in Figure 28 (which shows all U.S. wind projects) that low-specific- power turbines have established a strong foothold across the nation and over a wide range of wind speeds.

Likewise, taller towers are also being deployed across a wider array of sites (Figure 29). That said, very tall towers (>100m) still tend to be most concentrated within the upper Midwest and Northeast regions, two regions known to have higher-than-average (i.e., greater increases in wind speed with height), which makes taller towers more economical.

26 Land-Based Wind Market Report

Sources: ACP, U.S. Wind Turbine Database, AWS Truepower, Berkeley Lab Figure 28: Location of low specific power turbine installations

Sources: ACP, U.S. Wind Turbine Database, AWS Truepower, Berkeley Lab Figure 29: Location of tall tower turbine installations

27 Land-Based Wind Market Report

Wind projects planned for the near future are poised to continue the trend of ever-taller turbines

FAA data on total proposed turbine heights (from ground to blade tip extended directly overhead) in permit applications are reported in Figure 30. Note that these data represent total turbine height or “tip height”—not hub height—and include the combined effect of both the tower and half the rotor. Figure 30 shows the average FAA tip height, along with the distribution, for actual 2020 installations as well as turbines under construction, in advanced development, pending, and proposed.25

Average tip heights for projects that came online in 2020 are 525 feet (160 meters), and seem destined to climb higher in the next few years, reaching more than 660 feet (200 meters) among the “proposed” turbines. Historically, 500 feet was considered somewhat of a ceiling, due to more-involved FAA permitting and approval processes for turbines above that height (though not shown, the average FAA tip height never topped 500 feet until 2019, when it averaged 501 feet). The tallest turbines in the permitting process—those with a tip height of at least 750 feet—are intended for West Texas and central Illinois, but turbines of at least 650 feet appear likely to be installed in every region of the United States (Figure 31).

% of turbines Average Total height (feet) 100% 700

80% 650 750−850 ft

650−750 ft 60% 600 550−650 ft

450−550 ft 40% 550 <450 ft

Average: 20% 500 right axis

0% 450 2020 projects Under const. Adv. dev. Pending Proposed Sources: ACP, FAA files, Berkeley Lab Figure 30. Total turbine heights proposed in FAA applications, over time

25 Turbine heights reported in FAA permit applications represent the maximum height and can differ from what is ultimately installed. Historically, however, the FAA permit datasets have strongly conformed to subsequent actual installations on average.

28 Land-Based Wind Market Report

Note: Figure includes FAA data on under-construction, advanced development, pending, and proposed turbines Sources: FAA Obstacle Evaluation / Airport Airspace Analysis files, AWS Truepower, ACP, Berkeley Lab Figure 31. Total turbine heights proposed in FAA applications, by location

In 2020, thirty-three wind projects were partially repowered, most of which now feature significantly larger rotors and lower specific power ratings

The trend of partial wind project repowering continued through 2020, and involved replacing major components of turbines with more-advanced technology to increase energy production, extend project life, and access favorable tax incentives. In 2020, 33 projects were partially repowered, involving 1,827 turbines that totaled 3,087 MW prior to repowering. Retrofitted turbines ranged in age from 8 to 18 years old; the median age was 12 years.

As shown in Figure 32—which focuses on partial repowering in 2020—the most common retrofit in 2020 was the replacement of shorter with longer blades, but capacity changes were common as well, sometimes in opposing directions: 6 projects decreased capacity, 20 increased capacity, and 7 were unchanged. Overall, the average capacity of these retrofitted projects only increased modestly. The average rotor diameter increase was 14 meters. No turbines were retrofitted with new towers and therefore hub heights did not change. With the relatively small change in capacity but the larger change in rotor diameter, these retrofits drove a significant decrease in average specific power, with 32 out of 33 projects experiencing a decrease.

29 Land-Based Wind Market Report

Sources: ACP, Berkeley Lab, turbine manufacturers Figure 32. Change in average physical specifications of turbines that were partially repowered in 2020 Finally, in 2020, portions of three projects (343 turbines totaling 120 MW) were decommissioned and replaced with 50 new turbines totaling 148 MW, including new towers, blades, and nacelles—“full” repowering as opposed to “partial.”

30 Land-Based Wind Market Report

5 Performance Trends The average capacity factor in 2020 exceeded 40% among wind projects built in recent years, and reached 36% on a fleet-wide basis

Following the previous discussion of technology trends, this chapter presents data from a compilation of project-level capacity factors.26 The full data sample consists of 971 wind projects built between 1998 and 2019 and totaling 92,111 MW. Excluded from this assessment are older projects installed prior to 1998. In addition, 36 projects totaling more than 3.2 GW that were either partially or fully repowered in 2020 are excluded from the 2020 capacity factor sample, given that they were at least partly offline during a portion of the year. Unless otherwise noted, all capacity factors in this chapter are reported on an as-observed and unadjusted basis (i.e., after any losses from curtailment, less-than-full availability, wake effects, ice or soil on blades, etc.). When looking at performance degradation over time, however, adjustments are made for inter- annual variability in the wind resource.

To start, Figure 33 shows both individual project and average capacity factors in 2020, broken out by commercial operation date.27 Projects built in 2020 are excluded, as full-year performance data are not yet available for those projects. From left to right, Figure 33 shows an increase in weighted-average 2020 capacity factors when moving from projects installed in the 1998–2001 period to those installed in the 2004–2005 period. Subsequent project vintages through 2011 show little if any improvement in average capacity factors recorded in 2020. This pattern of stagnation is broken by projects installed in 2012–2013, and even more so by those that achieved commercial operations in 2014–2019.28

The average 2020 capacity factor among projects built from 2014 to 2019 was 41.4%, compared to an average of 29% among all projects built from 2004 to 2011, and 25.2% among all projects built from 1998 to 2001. Cumulative, fleet-wide performance has also increased over time, growing from under 27% in 1999 (not shown) to 36% in 2020 (shown in Figure 33). The improvement in capacity factor among more-recently built projects is impacted by several factors that are explored later, including project location and the quality of the wind resource at each site, turbine scaling and design, and performance degradation over time.

26 Capacity factor is a measure of the actual energy generated by a project over a given timeframe (typically annually) relative to the maximum possible amount of energy that could have been generated over that same timeframe if the project had been operating at full capacity the entire time. 27 Focusing on capacity factors in a single year, 2020, controls (at least loosely) for time-varying influences such as the degree of wind power curtailment or inter-annual variability in the strength of the wind resource. But it also means that the absolute capacity factors shown in Figure 33 may not be representative over longer terms if 2020 was not a representative year in terms of curtailment or the strength of the wind resource (though, as noted later, 2020 was a fairly average wind year overall). 28 The 2020 capacity factor of projects that were built in 2019 may be biased low, due to possible first-year “teething” issues, as projects may take a few months to achieve normal, steady-state production after first achieving commercial operations.

31 Land-Based Wind Market Report

Capacity Factor in 2020 60% Fleet-wide average Individual projects (by COD year) (across all COD years) 50%

40%

30%

20%

10% Generation-weighted average (by COD year) 0% 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019

1998-99 2000-01 2002-03 2004-05 Commercial Operation Year (COD Year) Sources: EIA, FERC, Berkeley Lab Figure 33. Calendar year 2020 capacity factors by commercial operation date

State and regional variations in capacity factors reflect the strength of the wind resource

The project-level spread in capacity factors shown in Figure 33 is enormous, with capacity factors in 2020 ranging from a minimum of 24% to a maximum of 56% among those projects built in 2019. Some of the spread—for projects built in 2019 and earlier—is attributable to regional variations in average wind resource quality. Figure 34 includes data on the full sample of projects built from 1998 through 2019 and also a subset of newer projects built from 2015 through 2019, and shows average state-level capacity factors in 2020. The overall range runs from 17%–45%, with considerably higher capacity factors in the interior of the country— where the wind resource is the strongest.

Note: States shaded in white have no projects in full sample (left) or in newer sample (right) Sources: EIA, FERC, Berkeley Lab Figure 34. Average calendar year 2020 capacity factor by state

32 Land-Based Wind Market Report

As shown earlier in Chapter 4, the rate of adoption of turbines with taller towers and lower specific power ratings varies somewhat by region. There is a greater preponderance of tall towers in the upper Midwest and Northeast regions than elsewhere, while lower specific power turbines are more-broadly spread across all regions. The relative degree to which projects in each region have employed these turbine design options also influences, to some extent, the capacity factors shown in Figure 34.

Turbine design and site characteristics influence performance, with declining specific power leading to sizable increases in capacity factor over the long term

The trends in average capacity factor by commercial operation date seen in Figure 33 can largely be explained by several underlying influences described in Chapter 4 and shown again in Figure 35. First, as documented in Chapter 4, there has been a trend toward lower specific power and higher hub heights. These two drivers are shown again in Figure 35 in index form, relative to projects built in 1998–1999 (with specific power shown in the inverse, to correlate with capacity factor movements). All else equal, a lower specific power will boost capacity factors, because there is more swept rotor area available (resulting in greater energy capture) for each watt of rated turbine capacity. Meanwhile, increasing turbine hub heights generally helps the rotor access higher wind speeds. Second, and potentially counterbalancing these drivers, there has been a tendency to build new wind projects in areas that feature lower average wind speeds,29 especially among projects installed from 2009 through 2012 as shown by the wind resource quality index in Figure 35. This trend reversed course in 2013 and 2014, but has since drifted lower once again (these wind resource trends are easier to see in Figure 27, where the y-axis scale is less-expansive). Finally, as shown later, two other drivers might include project age (given the possible degradation in performance among older projects) and increasing curtailment over the past few years (curtailment is baked into the capacity factors shown throughout this chapter).

Average Capacity Factor in 2020 Index of Capacity Factor Influences (1998−99=100) 50% 200 Inverse of built specific power 40% 175 Capacity factor

30% 150

Built turbine hub height 20% 125

10% Built wind resource quality at 80m 100

0% 75 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

1998-99 2000-01 2002-03 2004-05 Commercial Operation Year (COD Year) Note: In order to have all three indices be directionally consistent with their influence on capacity factor, this figure indexes the inverse of specific power (i.e., a decline in specific power causes the index to increase rather than decrease). Sources: EIA, FERC, Berkeley Lab Figure 35. 2020 capacity factors and various drivers by commercial operation date

29 As described earlier relating to Figure 27 (with further details in the Appendix), estimates of wind resource quality are based on site estimates of gross capacity factor at 80 meters, as derived from nationwide wind resource maps created for NREL by AWS Truepower. Those site estimates are indexed to projects built in 1998–1999.

33 Land-Based Wind Market Report

In Figure 35, the significant improvement in average 2020 capacity factors from among those projects built in 1998–2001 to those built in 2004–2005 is driven by both an increase in hub height and a decline in specific power, despite a shift toward somewhat lower-quality wind resource sites. The stagnation in average capacity factors that subsequently persists through 2011-vintage projects reflects relatively flat trends in both hub height and specific power, coupled with an ongoing decline in wind resource quality at built sites. Finally, the sharp increase in average capacity factors among projects built after 2011 is driven by a steep reduction in average specific power coupled with a marked improvement in the quality of wind resource sites. (Average hub height increased modestly over this period.) Looking ahead to 2021, projects with commercial operation dates in 2020 could possibly record lower capacity factors on average than those built in 2019, in light of a slight increase in average specific power coupled with steady hub height and a slight decline in average site quality.

To help disentangle the primary and sometimes competing influences of turbine design evolution and wind resource quality on capacity factor, Figure 36 controls for each. Across the x-axis, projects are grouped into four different categories, depending on the wind resource quality estimated for each site. Within each wind resource category, projects are further differentiated by their specific power. As would be expected, projects sited in higher wind speed areas generally realized higher capacity factors in 2020 than those in lower wind speed areas, regardless of specific power. Likewise, within each of the four wind resource categories, projects that fall into a lower specific power range realized significantly higher capacity factors in 2020 than those in a higher specific power range.

Average Capacity Factor in 2020 (1998−2019 projects) Specific Power bins (W/m2): 50% <250 250 -300 40% 300 350 -350 -400 30% ≥400

20%

10%

0% Lower Medium Higher Highest Estimated Wind Resource Quality at Site Note: The Appendix provides details on how the wind resource quality at each individual project site is estimated. Sources: EIA, FERC, Berkeley Lab Figure 36. Calendar year 2020 capacity factors by wind resource quality and specific power: 1998-2019 projects

Wind curtailment can differentially impact project performance across sites and regions

Curtailment of wind project output results from transmission inadequacy and other forms of grid and generator inflexibility in concert with wind over-supply. For example, over-generation can occur when wind generation is high but transmission capacity is insufficient to move excess generation to other load centers, or thermal generators cannot feasibly ramp down any further or quickly enough. This can push local wholesale power prices negative, thereby potentially triggering wind curtailment especially among projects not earning the PTC.

Curtailment might be expected to increase as wind energy penetrations rise, though—as shown in Figure 37— this has not always been the case. SPP has the highest wind penetration rate of any of the ISOs, yet its

34 Land-Based Wind Market Report curtailment rate in 2020 (at 2.5%) was lower than in MISO (5.0%) and ERCOT (4.6%). Moreover, in areas where curtailment has been particularly acute in the past, steps taken to address the issue have borne fruit. For example, Figure 37 shows that just 0.5% of potential wind energy generation within ERCOT was curtailed in 2014, down sharply from 17% in 2009. This decline in curtailment corresponds to a significant build-out of new transmission serving West Texas, most of which was completed by the end of 2013. Since 2014, however, wind penetration has continued to increase in ERCOT, and so too has wind curtailment.

Wind Curtailment Rate Wind Penetration (as a % of load) 18% 36% 2020 Curtailment 6% 15% 30% Wind Curtailment 4% 12% 24% 2%

9% 0% 18% Wind Penetration PJM SPP MISO NYISO CAISO ISO-NE 6% ERCOT 12%

3% 6%

0% 0% 2014-20 2007-20 2009-20 2015-20 2012-20 2014-20 2012-20 2007-20 SPP ERCOT MISO CAISO NYISO ISO-NE PJM Total

Sources: ERCOT, MISO, CAISO, NYISO, PJM, ISO-NE, SPP Figure 37. Wind curtailment and penetration rates by ISO Outside of MISO, ERCOT, and SPP, curtailment percentages in 2020 were relatively low: 1.5% in ISO-NE, 1.4% in NYISO, 0.6% in CAISO, and 0.1% in PJM. The overall wind power curtailment rate in 2020 across all seven regions was 3.4%, up from a low of 2.1% in 2016.

Temporal variations in wind speed also impact performance

The strength of the wind resource varies from year to year; moreover, the degree of inter-annual variation differs from site to site (and, hence, also region to region). This temporal and spatial variation, in turn, impacts project performance from year to year. Figure 38 shows national and regional indices of the historical inter- annual variability in the wind resource among the U.S. fleet over time.30 Though inter-annual variation has, at times, exceeded +/-20% at the regional level, geographical averaging has enabled nationwide variation to remain within +/-10%. In 2020, the national wind index was slightly above its long-term average, driven in part by higher wind speeds in the non-ISO West region.

30 These indices estimate changes in the strength of the average region- or fleet-wide wind resource from year to year and are constructed from ERA5 reanalysis wind speed data for individual project locations by applying applicable wind turbine power curves and then aggregating up to the region or fleet level (see the Appendix for more details). Note that these indices of inter- annual variability differ from the AWS Truepower wind resource quality data presented elsewhere, in that the former show variability from year to year across the entire region or fleet, while the latter focus on the multi-year long-term average wind resource at specific wind project sites.

35 Land-Based Wind Market Report

Average Annual Wind Resource Indices (Long-Term Average = 1.0) 1.2

1.1

1.0

0.9

0.8 CAISO West (non-ISO) SPP MISO ERCOT PJM NYISO ISO-NE National 0.7 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 Sources: ERA, Berkeley Lab; methodology behind the index of inter-annual variability is explained in the Appendix Figure 38. Inter-annual variability in the wind resource by region and nationally

Wind project performance degradation also explains why older projects did not perform as well in 2020

A final variable that could influence the improvement in 2020 capacity factors among more recent projects is project age. If wind turbine (and project) performance tends to degrade over time, then older projects—e.g., those built from 1998 to 2001—may have performed worse in 2020 than more recent projects simply due to their relative age. Figure 39 explores this question by graphing median (and 25th to 75th percentile ranges) “-normalized” (i.e., correcting for inter-annual variability in the strength of the wind resource) capacity factors over time. Here, time is defined as the number of full calendar years after each individual project’s commercial operation date, and each project’s capacity factor is indexed to 100% in year two in order to focus solely on changes in capacity factor over time, rather than on absolute capacity factor values. Year two is chosen as the index base to reflect the initial production ramp-up period commonly experienced by wind projects as their operators work through and resolve initial “teething” issues during the first year of operations.

Figure 39 suggests some amount of performance decline, especially among projects built before 2008. Projects built in 2008 and later appear, on average, to have relatively stable output for the first ten years, but then experience a decline in years 11 and 12—perhaps reflecting a change in how projects are operated once they age beyond the 10-year PTC window. Hamilton et al. (2020) explore these performance trends in more depth.

36 Land-Based Wind Market Report

Indexed Capacity Factor (Year 2 = 100%)

110% Newer (post-2007) projects (with 25th and 75th percentile range) 100%

90%

80%

70% Older (pre-2008) projects (with 25th and 75th precentile range) 60%

50% 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Year of Operations Sources: EIA, FERC, Berkeley Lab Figure 39. Changes in project-level capacity factors as projects age Taken together, Figure 33 through Figure 39 suggest that, in order to understand trends in empirical capacity factors, one needs to consider (and ideally control for) a variety of parameters. These include not only wind power curtailment and the evolution in turbine design, but also a variety of spatial and temporal wind resource considerations—such as the quality of the wind resource where projects are located, inter-year wind resource variability, and even project age.

37 Land-Based Wind Market Report

6 Cost Trends Wind turbine prices remained well below levels seen a decade ago

Wind turbine prices have dropped substantially since 2008, despite continued technological advancements that have yielded increases in hub heights and especially rotor diameters. Prices largely held steady in 2020.

Figure 40 depicts wind turbine transaction prices from a variety of sources: (1) Vestas, SGRE, and Nordex, on those companies’ global average turbine pricing, as reported in corporate financial reports; (2) BloombergNEF (2020) and Wood Mackenzie (2021a), on those companies’ turbine price indices by contract signing date; and (3) 121 U.S. wind turbine transactions announced from 1997 through 2016, as previously collected by Berkeley Lab. Wind turbine transactions can differ in the services included (e.g., whether towers are provided, the length of the service agreement, etc.), turbine characteristics (and therefore performance), and the timing of future turbine delivery. These differences drive some of the observed intra-year variability in transaction prices. Most of the prices and transactions reported in the figure are inclusive of towers and delivery to the site.

Turbine Price (2020 $/kW)

2,500 U.S. turbine orders 1,000 BNEF global index 800 Wood Mackenzie U.S. index 600 2,000 Nordex global average SGRE global average 400 M Vestas global average 200 - Nordex SGRE BNEF Vestas W 0 1,500

1,000

500

0 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021

Sources: Berkeley Lab, annual financial reports, forecast providers Figure 40. Reported wind turbine transaction prices over time After hitting an initial low of roughly $900/kW from 2000 to 2002, average wind turbine prices increased by approximately $900/kW through 2008, rising to an average of $1,800/kW. This increase in turbine prices was caused by several factors, including a decline in the value of the U.S. dollar relative to the Euro; increased materials, energy, and labor input prices; a general increase in turbine manufacturer profitability due in part to strong demand growth; and increased costs for turbine warranty provisions (Moné et al. 2017).

Since 2008, wind turbine prices have declined by more than 50%, reflecting a reversal of some of the previously mentioned underlying trends that had earlier pushed prices higher (Moné et al. 2017) as well as increased competition among manufacturers and significant cost-cutting measures on the part of turbine and component suppliers. Data signal recent average pricing in the range of $775/kW to $850/kW. This decline in turbine prices has been coupled with improved turbine technology (e.g., the recent growth in average hub heights and rotor diameters). The turbine price trends have exerted downward pressure on total project costs,

38 Land-Based Wind Market Report whereas increased rotor diameters and hub heights have boosted capacity factors—both factors yielding lower wind power prices and levelized cost of energy.

Lower turbine prices have driven reductions in reported installed project costs

Berkeley Lab also compiles data on the total installed cost of wind projects in the United States, including data on 65 projects completed in 2020 totaling 12,173 MW, or 72% of the wind power capacity installed in that year. In aggregate, the dataset includes 1,128 completed wind power projects in the continental United States totaling 106,996 MW and equaling roughly 88% of all wind power capacity installed as of the end of 2020. In general, reported project costs reflect turbine purchase and installation, balance of plant, and any substation and/or interconnection expenses. Data sources are diverse, however, and are not all of equal credibility, so emphasis should be placed on overall trends in the data rather than on individual project-level estimates.

As shown in Figure 41, the average installed costs of projects declined from the beginning of the U.S. wind industry in the 1980s through the early 2000s, and then increased—reflecting turbine price changes—through the latter part of the last decade. Costs peaked in 2009–2010. Though project-level costs have declined since 2010, they have largely held steady over the last few years.

Installed Project Cost (2020 $/kW) 6,000

5,000

4,000 Capacity-weighted average 3,000

2,000

1,000

0 1985 1990 1995 2000 2005 2010 2015 2020

Commercial Operation Date Note: Area of “bubble” is proportional to project capacity Sources: Berkeley Lab, EIA (some data points suppressed to protect confidentiality) Figure 41. Installed wind power project costs over time In 2020, the capacity-weighted average installed project cost within the sample stood at $1,460/kW. This is down by more than 40% from the average reported costs in 2009 and 2010, but is roughly on par with the installed costs experienced in the early 2000s.

Installed costs differed by region

Regional differences in average project costs are also apparent and may occur due to variations in labor costs, development costs, transportation costs, siting and permitting requirements and timeframes, and other balance- of-plant and construction expenditures—as well as variations in average project size and the turbines deployed in different regions (e.g., use of low-wind-speed technology in regions with lesser wind resources).

39 Land-Based Wind Market Report

Figure 42 presents capacity-weighted average installed costs over the previous five years, by region. Figure 43 presents data only from the latest year—2020. Sample size is limited in some regions and years. However, costs have largely declined over the last five years. The lowest-cost projects in 2020 were located in ERCOT and the Western states (excluding California), with average costs of $1,280/kW and $1,380/kW respectively. SPP and MISO—historically two low-cost regions—saw average costs rise over the last year or two.

Average Cost (2020 $/kW) National ERCOT West SPP PJM MISO ISO-NE CAISO 2,800

2,400

2,000

1,600

1,200

800

400

0 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020

Note: Data for NYISO and the Southeast are not available over this period. Other regions are missing data for specific years. Sources: Berkeley Lab, EIA Figure 42. Installed wind power project costs by region, over time

Installed Cost of 2020 Projects (2020 $/kW)

2,000

1,500

1,000

500

$1,283 $1,378 $1,419 $1,471 $1,633 0 ERCOT West (non-ISO) PJM SPP MISO Note: Size of bubble reflects project capacity. Other regions lack adequate data for inclusion. Sources: Berkeley Lab Figure 43. Installed wind power project costs by region, in 2020

40 Land-Based Wind Market Report

Installed costs differed by project size

Installed costs exhibit economies of scale, which are especially evident when moving from small- to medium- sized projects. Figure 44 shows that among the sample of projects installed in 2019 and 2020, there is a substantial drop in per-kW average installed costs when moving from projects of 5 MW or less to projects in the 20–50 MW and 50–100 MW ranges. Economies of scale continue, though to a lesser degree, as project size increases beyond these thresholds.

Installed Project Cost (2020 $/kW)

3,000 Capacity-weighted average Individual projects

2,000

1,000

0 ≤5 MW 5-20 MW 20-50 MW 50-100 MW 100-200 MW >200 MW Project Capacity Source: Berkeley Lab Figure 44. Installed wind power project costs by project size: 2019 and 2020 projects

Operations and maintenance costs varied by project age and commercial operations date

Operations and maintenance (O&M) costs are an important component of the overall cost of wind energy and can vary substantially among projects. Unfortunately, publicly available market data on actual project-level O&M costs are not widely available. Even where data are available, care must be taken in extrapolating historical O&M costs given the dramatic changes in wind turbine technology that have occurred over time (see Chapter 4).

Berkeley Lab has compiled limited O&M cost data for 188 installed wind power projects, totaling 19,366 MW and with commercial operation dates of 1982 through 2019. These data cover facilities owned by both IPPs and utilities, although data since 2004 are exclusively from utility-owned projects and so may not be broadly representative. A full time series of O&M cost data, by year, is available for only a small number of projects; in all other cases, O&M data are available for just a subset of years of project operations. Although not all data sources clearly define what items are included in O&M costs, in most cases the reported values include the costs of wages and materials associated with operating and maintaining the wind project, as well as rent.31 Other ongoing expenses, including general and administrative expenses, taxes, property insurance,

31 The vast majority of the recent data derive from FERC Form 1, which uses the Uniform System of Accounts to define what should be reported under “operating expenses”—namely, those operational costs associated with supervision and engineering, maintenance, rents, and training. Though not entirely clear, there does appear to be some leeway within the Uniform System of Accounts for project owners to capitalize certain replacement costs for turbines and turbine components and report them under “electric plant” accounts rather than maintenance accounts.

41 Land-Based Wind Market Report depreciation, and workers’ compensation insurance are generally not included. As such, Figure 45 and Figure 46 are not representative of total operating expenses for wind power projects.

Figure 45 shows O&M costs by commercial operation date.32 Here, each project’s O&M costs are depicted as average annual O&M costs from 2000 through 2020, based on however many years of data are available for that period. For example, for projects that reached commercial operation in 2019, only 2020 data are available, and that is what is shown. Many other projects only have data for a subset of years, so each data point in the chart may represent a different averaging period within the overall 2000–2020 period. The chart highlights the 89 projects, totaling 14,418 MW, for which 2020 O&M cost data were available; those projects have either been updated or added to the chart since the previous edition of this report.

Average Annual O&M Cost, 2000−2020 (2020 $/kW-yr) 160 Projects with no 2020 O&M data Projects with 2020 O&M data

120

80

40

0 1984 1989 1994 1999 2004 2009 2014 2019 Commercial Operation Date Source: Berkeley Lab; some data points suppressed to protect confidentiality Figure 45. Average O&M costs for available data years from 2000 to 2020, by commercial operation date The data demonstrate that O&M costs are far from uniform across projects. Figure 45 also suggests that projects installed in the past decade have, on average, incurred lower O&M costs than those installed earlier. Specifically, capacity-weighted average 2000–2020 O&M costs for the 24 projects in the sample constructed in the 1980s equal $74/kW-year, dropping to $62/kW-year for the 37 projects installed in the 1990s, to $27/ kW-year for the 65 projects installed in the 2000s, and $25/kW-year for the 59 projects installed since 2010. This decline may be due to at least two factors: (1) O&M costs generally increase as turbines age and component failures become more common; and (2) projects installed more recently, with larger and more mature turbines and more sophisticated O&M practices, may experience lower overall O&M costs.

Limitations in the underlying data do not permit the influence of these two factors to be clearly distinguished. Nonetheless, to help illustrate key trends, Figure 46 shows median annual O&M costs over time, based on project age (i.e., the number of years since the commercial operation date) and segmented into three project- vintage groupings. Though sample size is limited, the data show an upward trend in project-level O&M costs as projects age, at least among the oldest projects in the sample. Figure 46 also shows that projects installed

32 For projects installed in multiple phases, the commercial operation date of the largest phase is used. For repowered projects, the date at which repowering was completed is used.

42 Land-Based Wind Market Report over the last 15 years have had, in general, lower O&M costs than those installed in the earlier years of 1998– 2005, at least for the first 14 years of operation.

Median Annual O&M Cost (2020 $/kW-year) 75 Commercial Operation Date: 1998-2005 60 2006-2011 2012-2019 45

30

15

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Project Age (Number of Years Since Commercial Operation Date) Source: Berkeley Lab; medians shown only for groups of two or more projects, and only projects >5 MW are included Figure 46. Median annual O&M costs by project age and commercial operation date As indicated previously, these data include only a subset of total operating expenses. A U.S. wind industry survey of total operating costs shows that these expenses for recently installed projects are anticipated to average between $33/kW-year and $59/kW-year, with a mid-point of ~$44/kW-year (Wiser et al. 2019). The disparity between these estimates of total operating costs and the costs reported in Figure 45 and Figure 46 reflects, in large part, differences in the scope of expenses reported; the survey noted that turbine O&M is expected to constitute less than half of total operating costs (Wiser et al. 2019).

43 Land-Based Wind Market Report

7 Power Sales Price and Levelized Cost Trends Wind power purchase agreement prices remain low

Earlier chapters documented trends in capacity factors, installed project costs, O&M costs, and project financing—all of which are determinants of the wind power purchase agreement (PPA) prices and levelized cost of energy (LCOE) estimates presented in this chapter.

Berkeley Lab collects data on wind PPA prices, resulting in a dataset that includes 506 PPAs totaling 50,465 MW from wind projects that have either been built or are planned for installation later in 2021 or beyond. All of these PPAs bundle together the sale of electricity, capacity, and renewable energy certificates (RECs; a later text box highlights REC prices), and most of them have a utility as the counterparty.33 Except where noted, PPA prices are expressed on a levelized basis over the full term of each contract and are reported in real 2020 dollars.34 Whenever individual PPA prices are averaged together, the average is generation-weighted. Whenever they are broken out by time, the date on (or year in) which the PPA was executed is used. Because PPA prices are reduced by the receipt of state and federal incentives and are influenced by various local policies and market characteristics, they do not directly represent wind energy generation costs. Accordingly, at the end of this chapter, the data presented earlier in this report are leveraged to estimate project-level and average wind LCOE for a large sample of U.S. wind projects.

Figure 47 plots contract-level levelized wind PPA prices by contract execution date, showing a clear decline in PPA prices since 2009–2010, both overall and by region. As a result of the low average project costs and high average capacity factors shown earlier in this report, ERCOT and SPP tend to be the lowest-priced regions. Of note, PPA prices have not smoothly declined over time. Instead, prices declined through 2003, then rose though 2009 with the increased turbine and installed costs presented earlier as well as with general price increases during this period in the power and natural gas markets. Following that rise was a steep reduction and, most recently, some stabilization or even slight increase in PPA prices.

33 Though some PPAs with corporate offtakers are included in the sample, in many cases such PPAs are synthetic or financial arrangements in which the project sponsor enters into a “contract for differences” with the corporate offtaker around an agreed- upon strike price. Because the strike price is not directly linked to the sale of electricity, it is rarely disclosed (at least through traditional sources, like regulatory filings). 34 Having full-term price data (i.e., pricing data for the full duration of each PPA, rather than just historical PPA prices) enables these PPA prices to be presented on a levelized basis (levelized over the full contract term), which provides a complete picture of wind power pricing (e.g., by capturing any escalation over the duration of the contract). Contract terms range from 5 to 35 years, with 20 years being by far the most common (at 55% of the sample; 87% of contracts in the sample are for terms ranging from 15 to 25 years). Prices are levelized using a 4% real discount rate.

44 Land-Based Wind Market Report

Levelized PPA Price (2020 $/MWh) CAISO West (non-ISO) 140 MISO SPP 120 ERCOT 100 PJM NYISO 80 Southeast (non-ISO) ISO-NE 60

40

20

0 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 PPA Execution Date Note: Size of bubble reflects contract capacity. Source: Berkeley Lab, FERC Figure 47. Levelized wind PPA prices by PPA execution date and region (full sample) Figure 48 provides a smoother look at the time trend nationwide and regionally by averaging the individual levelized PPA prices shown in Figure 47, and consolidating the regional breakdown into just three categories: West, Central, and East. After topping out above $70/MWh for PPAs executed in 2009, the national average levelized price of wind PPAs within the Berkeley Lab sample has dropped. In the Central region of the country, recent pricing is below $20/MWh. In the West and East, prices tend to average roughly $30/MWh.

Average Levelized PPA Price (2020 $/MWh) 100 East West 80

60

40

20 Central Nationwide 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

1996-99 2000-01 2002-03 2004-05 PPA Execution Year 2019-20 Note: West = CAISO, West (non-ISO); Central = MISO, SPP, ERCOT; East = PJM, NYISO, ISO-NE, Southeast (non-ISO) Source: Berkeley Lab, FERC Figure 48. Generation-weighted average levelized wind PPA prices by PPA execution date and region

45 Land-Based Wind Market Report

The trend of rising PPA prices from 2003 to 2009 and lower prices since then is directionally consistent with the turbine price and installed project cost trends shown earlier in Chapter 6. In addition, the turbine scaling described in Chapter 4 has, on average, boosted the capacity factors of more recent projects, as documented in Chapter 5. Scaling has also enabled reductions in operating costs, as described in Chapter 6. This combination of declining CapEx and OpEx and improved performance—along with low interest rates (as shown earlier in Figure 21)—has driven wind PPA prices to today’s low levels.

Recent wind power purchase agreements are priced in the mid-teens in some cases

Figure 49 narrowly focuses on just the most recent PPAs in the sample, maintaining the three larger regions introduced in Figure 48. This sample of PPAs signed since the start of 2014 includes 135 contracts totaling 17,333 MW. The levelized prices of these 135 PPAs range from $8.6/MWh to $82.7/MWh. Contract terms range from 10 to 35 years, with an average of 20 years.

Levelized PPA Price (2020 $/MWh) West 80 Central 70 East 60

50

40

30

20

10

0 2014 2015 2016 2017 2018 2019 2020 2021 PPA Execution Date Note: Size of bubble reflects contract capacity. Source: Berkeley Lab, FERC Figure 49. Levelized wind PPA prices by PPA execution date and region (recent sample) LevelTen Energy’s PPA price indices confirm low but rising PPA prices, and regional variations

In contrast to the PPAs summarized above, which principally involve utility purchasers, LevelTen Energy (2021) provides an index of wind PPA offers made to large, end-use customers.

Each quarter, the LevelTen Energy PPA Price Index reports the prices that wind and solar developers have offered for PPAs available on the LevelTen Marketplace. Contract terms tend to range from 10-15 years, reflective of the shorter terms typically pursued by end-use customers that purchase wind energy relative to the utility PPAs summarized earlier. Price data are aggregated and reported on a ‘P25’ basis, referring to the most competitive 25th percentile of offer prices.

As shown in Figure 50, prices in specific regions have generally risen over the last six quarters, and vary by ISO. Among regions reporting data, CAISO features the highest wind PPA pricing (~$50/MWh), whereas the lowest prices are found in ERCOT and SPP (~$20/MWh).

46 Land-Based Wind Market Report

Level10 PPA Price Index (nominal $/MWh, 25th percentile of offers)

CAISO 50

40 PJM

30 MISO SPP 20

ERCOT 10

0 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 2018 2019 2020 2021

Source: LevelTen Energy

Figure 50. LevelTen Energy wind PPA price index by quarter of offer The (unsubsidized) average levelized cost of wind energy has fallen to around $33/MWh

In a competitive market, long-term PPA prices can be thought of as reflecting the LCOE reduced by the value of any incentives received (e.g., the PTC). Hence, as a first-order approximation, LCOE can be estimated simply by adding the levelized value of incentives received to the levelized PPA prices. LCOE can also be estimated more directly from its components, and Berkeley Lab has data on both the installed cost and capacity factor of 104 GW of wind power projects installed from 1998 through 2020, representing 85% of all capacity built over that period. Here, those data are used, in conjunction with estimates of operational costs, financing costs, project life and other factors, to estimate LCOE in real 2020 dollars (see the Appendix for details on the data and calculations). One benefit of this “bottom up” approach to estimating LCOE is that it relies on a large sample of project-level installed cost and performance data, covering more projects than the PPA sample.

Figure 51 depicts the resulting average LCOE values over time on a national basis. As shown, average wind LCOE declined from $102/MWh in 1988−1999 to $67/MWh in 2004−2005, before rising to $93/MWh in 2009. Subsequently, average LCOE declined rapidly through 2018, to $33/MWh. The national average LCOE of newly built wind projects has largely held steady since 2018, and stood again at $33/MWh in 2020.

47 Land-Based Wind Market Report

Average and Project-level LCOE (2020 $/MWh) 160

140

120

100

80

60

40

20 102 86 80 67 79 79 87 93 90 89 75 60 50 45 42 39 33 34 33 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2019 2020 2017 2018

1998-99 2000-01 2002-03 2004-05 Commercial Operation Year Note: Size of bubble reflects project capacity.

Source: Berkeley Lab

Figure 51. Estimated levelized cost of wind energy by commercial operation date Levelized costs vary by region, with the lowest costs in ERCOT, SPP, and the non-ISO West

Regional LCOE estimates span a wide range, and sample size is small in some regions and years. Nonetheless, the lowest average LCOEs for projects built in 2020—only considering regions with a larger sample—are found in ERCOT ($29/MWh), SPP ($32/MWh), non-ISO West ($32/MWh), and MISO ($35/MWh).

National ERCOT SPP West PJM ISO-NE CAISO NYISO 2020 Avg = MISO $33/MWh $29/MWh $32/MWh $32/MWh $35/MWh $39/MWh $60/MWh no data no data 90

60 LCOE (2020 $/MWh) 30 Average

0 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 2016 2018 2020 Source: Berkeley Lab

Figure 52. Estimated levelized cost of wind energy, by region

48 Land-Based Wind Market Report

Renewable Energy Certificate (REC) Prices

Wind power sales prices presented in this report reflect bundled sales of both electricity and RECs. Projects that sell RECs separately from electricity, thereby generating two sources of revenue, are excluded. REC markets are fragmented, but consist of two distinct segments: compliance markets, in which RECs are purchased to meet state RPS obligations, and green power markets, in which RECs are purchased on a voluntary basis. Mandatory RPS programs exist in 30 states and Washington, D.C. In recent years, roughly one-third of these states have increased their RPS targets, in many cases to levels ranging from 50% to 100% of retail electricity sales. Voluntary markets for renewable energy have also grown.

The figure below presents indicative data of spot-market REC prices in both compliance and voluntary markets. Clearly, spot REC prices have varied substantially, both over time and across states, though prices within regional power markets (New England and PJM) are linked to varying degrees.

REC prices in most compliance markets rose in 2020, reflecting tightening supplies relative to current RPS demand. In New England, REC prices (outside of ME) reached a 5-year high in 2020, ending the year at roughly $43/MWh, following a rapid rise in 2019. In PJM, REC prices (outside of DC) continued on their slow upward trajectory of the past several years, ending the year just above $10/MWh. Prices for RECs offered in the national voluntary market and for RPS compliance in Texas, while well below levels in most other markets, roughly doubled in value over 2020, ending the year at $1.6/MWh.

New England PJM Texas & Voluntary Market 80 40 8 CT DC TX MA DE Vol. (nat'l) ME IL Vol. (west) 60 NH 30 MD 6 RI NJ OH PA 40 20 4 2020 $/MWh 20 10 2

0 0 0 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 2010 2012 2014 2016 2018 2020 Notes: Data for compliance markets focus on “Class I” or “Tier I” RPS requirements; these are the requirements for more-preferred resource types or vintages and are therefore the markets in which wind would typically participate. Plotted values are the monthly averages of daily closing prices for REC vintages from the current or nearest future year traded. Source: Marex Spectron

49 Land-Based Wind Market Report

8 Cost and Value Comparisons Despite low PPA prices, wind faces competition from solar and gas

Figure 53 plots wind PPA prices against utility-scale solar PPA prices on a levelized basis since 2009 (the dashed blue and gold lines show the generation-weighted average wind and solar PPA prices in each year, respectively). Although the gap between wind and solar PPA prices was quite wide a decade ago, that gap has narrowed considerably in recent years, as solar prices have fallen more rapidly than wind prices.35

The figure also shows that wind PPA prices—and, more recently, utility-scale solar PPA prices—have, in many cases, been competitive with the projected fuel costs of gas-fired combined cycle generators. Specifically, the black dash markers show the 20-year levelized fuel costs—converted from natural gas to power terms at an assumed heat rate of 7.5 million British Thermal Units (MMBtu) per MWh—from then- current EIA projections of natural gas prices delivered to electricity generators.36 Supported by federal tax incentives, the average levelized wind and solar PPA prices within this contract sample have, for several years now, been below the projected levelized cost of burning natural gas in existing gas-fired combined cycle units.

Levelized PPA and Gas Price (2020 $/MWh) 180 PV PPA prices 160 140 120 Levelized 20-year EIA gas price projections 100 80 60 40 20 Wind PPA prices 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 PPA Execution Date and Gas Projection Year Note: Smallest bubble sizes reflect smallest-volume PPAs (<5 MW), whereas largest reflect largest-volume PPAs (400 MW) Sources: Berkeley Lab, FERC, EIA Figure 53. Levelized wind and solar PPA prices and levelized gas price projections Rather than levelizing the wind PPA prices and gas price projections, Figure 54 plots the future stream of wind PPA prices (the 10th, 50th, and 90th percentile prices are shown) from PPAs executed in 2018–2020 against the EIA’s latest projections of just the fuel costs of natural gas-fired generation.37 As shown, median wind PPA

35 The solar PPA prices are sourced from Berkeley Lab’s “Utility-Scale Solar” data series. 36 For example, the black dash marker in 2009 shows the 20-year levelized gas price projection from Annual Energy Outlook 2009, while the black dash in 2021 shows the same from Annual Energy Outlook 2021 (both converted to $/MWh terms at a constant heat rate of 7.5 MMBtu/MWh). 37 The fuel cost projections come from the EIA’s Annual Energy Outlook 2021 publication. The upper and lower bounds of the fuel cost range reflect the low (and high, respectively) oil and gas resource and technology cases. All fuel prices are converted from $/MMBtu into $/MWh using the heat rates implied by the modeling output (which start at roughly 7.8 MMBtu/MWh in 2021 and gradually decline to roughly 7.2 MMBtu/MWh by 2040).

50 Land-Based Wind Market Report prices from contracts executed in the past three years are consistently below the low end of the projected natural gas fuel cost range, while the 90th percentile wind PPA prices are initially above the high end of the fuel cost range, but fall within the overall range by 2030. Wind PPA pricing declines over time, in real 2020$.

2020 $/MWh 60 Range of AEO2021 natural gas fuel cost projections 50 AEO2021 reference case natural gas fuel cost projection 40

Gas 30

20 Wind

10 Median wind PPA price (with 10th and 90th percentile range) 0 2021 2025 2030 2035 2040

Sources: Berkeley Lab, FERC, EIA Figure 54. Wind PPA prices and natural gas fuel cost projections by calendar year over time Figure 54 also hints at the long-term value that wind power might provide as a “hedge” against rising and/or uncertain natural gas prices. The wind PPA prices that are shown have been contractually locked in, whereas the fuel cost projections to which they are compared are highly uncertain. Actual fuel costs could ultimately be lower or higher. Either way, as evidenced by the widening range of fuel cost projections over time, it becomes increasingly difficult to forecast fuel costs with any accuracy as the term of the forecast increases. Wind PPA prices have been broadly attractive compared to wind’s grid-system value in wholesale power markets

In many regions of the country, wind projects participate in organized wholesale electricity markets. In some cases, wind projects directly bid into those markets, and earn the prevailing market price. In other cases— especially when a PPA is in place—the wind purchaser will schedule the wind energy into the market, paying the wind project owner the pre-negotiated PPA price but earning revenue from the prevailing wholesale market price. PPAs between wind generators and commercial customers are often a hybrid of these two models.

In all of these cases, the revenue earned (or that could have been earned) from the sale of wind into wholesale markets is reflective of the market value of that generation from the perspective of the electricity system. In the case of merchant wind projects, the link is direct and affects the revenue of the plant. In the case of wind projects sold under a PPA, on the other hand, the pre-negotiated PPA price establishes plant revenue and, depending of the specifics of the PPA, pricing may or may not be linked to wholesale market prices. In this latter case, however, the revenue earned or that would have been earned by the sale of wind in the wholesale market still reflects the underlying market value of that wind—but in this instance, for the purchaser, in the form of an avoided cost. This is because wholesale electricity prices reflect the timing of when energy is cheap or expensive and embed the cost of transmission congestion and losses. A purchaser could, in theory, obtain

51 Land-Based Wind Market Report power from the wholesale market instead of from a wind project. A wind project’s estimated revenue were it participating in the wholesale market therefore reflects costs avoided by the purchaser of wind under a PPA.

This (potential) revenue—or value—can be segmented into “energy” market value and, where capacity markets or requirements exist, “capacity” value. Wholesale energy prices vary over time, and by location. They are strongly influenced by the cost of natural gas. Because wind power deployment is sometimes concentrated in areas with limited transmission capacity, wholesale energy prices at the local pricing nodes to which wind plants interconnect are often suppressed and the relationship to the cost of natural gas is diminished. Even absent transmission constraints, wind plants push wholesale energy prices lower when wind output is high. More generally, the temporal profile of wind output is not always well-aligned with customer load and system needs, potentially further reducing the energy market value of wind generation. Some of these tendencies also apply to wind’s capacity value, which is impacted by the cost of capacity but also by regional rules that define the credit that wind receives for providing capacity.

Figure 55 estimates the historical wholesale energy and capacity market value of wind across a number of different regions of the country. Specifically, the energy market value of wind is estimated using plant-level hourly wind output profiles and real-time hourly wholesale energy pricing patterns at the nearest pricing node (i.e., locational marginal prices, LMPs). Plant-level capacity values are estimated based on the relevant capacity price or cost for the region in question, and local rules for wind’s capacity credit.38 Energy and capacity are summed for each plant, and plant-level total value estimates are then averaged to estimate regional values. As a result, the analysis considers the output profile of wind, the location of wind, and how those characteristics interact with local wholesale energy and capacity prices and rules, ultimately yielding an estimate of the revenue that would have been earned had wind sold its output at the hourly LMP and also considering any possible capacity-based revenue. The figure then contrasts those wholesale market value estimates for wind with nationwide generation-weighted average levelized wind PPA prices (with error bars denoting the 10th and 90th percentiles) based on the years in which the PPAs were executed. The comparison between market value estimates and PPA prices is relevant in as much as PPA prices reflect the cost of wind to the purchaser, whereas wholesale market value reflects a portion of the value of that wind generation.

These estimates show that the wholesale market value of wind has generally declined over the last 12 years and varies by region. With the sharp drop in wholesale prices and therefore market value of wind in 2009, average wind PPA prices tended to well-exceed the wholesale market value of wind from 2009 to 2012. With continued declines in PPA prices, however, those prices reconnected with the market value of wind in 2013 and have remained generally in competitive territory in subsequent years. This suggests that—with the help of the PTC, which reduces PPA prices—wind developers and offtakers are successfully contracting at levels that are generally comparable in terms of both cost and value. Over the last couple years, natural gas prices have led wholesale prices lower and, among this sample, wind PPA prices have held steady or even risen. In 2020, natural gas and wholesale electricity prices hit new lows, in part a result of the economic response to the pandemic. Natural gas prices have since risen substantially above 2020 levels, and in mid-2021 are averaging higher than 2019 levels. This indicates that electricity prices and wind value will likely rebound in 2021.

38 The Appendix provides additional details on the methods used to estimate the wholesale energy and capacity value of wind.

52 Land-Based Wind Market Report

Wholesale Market Value and PPA Prices (2020 $/MWh) 120 CAISO Average Levelized Wind PPA Price with 10th/90th 100 Percentiles (by year of PPA execution) ISO-NE

80 PJM

SPP 60 NYISO 40 MISO

20 ERCOT

0 2008 2010 2012 2014 2016 2018 2020 Note: Hourly wind output profiles and wholesale prices are not available for all historical years for all regions. Sources: Berkeley Lab, ABB, ISOs Figure 55. Regional wholesale market value of wind and average levelized long-term wind PPA prices over time

Because many of the regional wholesale market value estimates are in a similar range, it is difficult to discern individual regional data points in Figure 55. Accordingly, Figure 56 presents these estimates of wind’s wholesale market value, by region, but only for the latest year—2020. The figure also disaggregates the market value estimates into their constituent parts: energy and capacity. The average market value of wind in 2020 was the lowest in ERCOT ($11/MWh), MISO ($11/MWh), NYISO ($13/MWh), and SPP ($14/MWh), whereas the higher-value markets were CAISO ($29/MWh), ISO-NE ($25/MWh), and PJM ($20/MWh). Energy value represented the largest share of the total, with capacity value varying widely regionally and being considerably lower in absolute magnitude.

53 Land-Based Wind Market Report

Wholesale Market Value in 2020 (2020 $/MWh) 30 capacity value

energy value 20

10

0 CAISO ISO-NE PJM SPP NYISO MISO ERCOT Sources: Berkeley Lab, ABB, ISOs Figure 56. Regional wholesale market value of wind in 2020, by region

Important Note on Price and Value Comparisons

Notwithstanding the comparisons made in this chapter, neither the wind prices nor wholesale market value estimates (nor fuel cost projections) reflect the full social costs of power generation and delivery. Among the various shortcomings of comparing wind (and solar) PPA prices with wholesale value and natural-gas cost estimates in this manner are the following: • Wind (and solar) PPA prices are reduced by federal and state incentives. Similarly, wholesale electricity prices (or fuel cost projections) are reduced by any financial incentives provided to thermal generation and its fuel production. Wholesale prices may also not fully account for the health and environmental costs of various generation technologies (though a later section within this chapter assesses the health and climate benefits of wind), and for other societal concerns such as fuel diversity and resilience. • Wind (and solar) PPA prices do not fully reflect integration, resource adequacy, or transmission costs, while wholesale electricity prices (or fuel cost projections) also do not fully reflect transmission costs, and may not fully reflect capital and fixed operating costs. • Wind and solar PPA prices—once established—are fixed and known. The estimated wholesale market value of wind represents historical values, whereas future natural gas prices are uncertain. Said another way, levelized wind (and solar) PPA prices represent a future stream of prices that has been locked in (and that often extends for 20 years or longer), whereas the wholesale value estimates are pertinent to just the specific historical years evaluated, and future natural gas prices reflect uncertain forecasts. In short, comparing levelized long-term wind PPA prices with either yearly estimates of the wholesale market value of wind or forecasts of the fuel costs of natural gas-fired generation is not appropriate if one’s goal is to account fully for the costs and benefits of wind energy relative to other generation sources. Nonetheless, these comparisons still provide some sense for the short-term competitive environment facing wind energy, and convey how those conditions have shifted over time. A subset of issues related to transmission and grid-integration of wind energy are covered in two text boxes below.

54 Land-Based Wind Market Report

The grid-system market value of wind varies by project location

Figure 57 presents the 2020 wind power market value estimates at a project level. These estimates span a narrower range than in past years, with the 10th, 50th, and 90th percentile values equaling $7, $15, and $29 per MWh, respectively.

The figure also shows some variability in market value within each region, with areas facing transmission congestion and high wind penetrations generally experiencing lower market value. Higher market value estimates are found in uncongested areas, areas with higher average wholesale prices, and areas where wind output profiles are more-correlated with electricity demand. (Developments related to new transmission and wind energy are discussion in an accompanying text box).

Sources: Berkeley Lab, ABB, ISOs Figure 57. Project-level wholesale market value of wind in 2020

The grid-system market value of wind tends to decline with wind penetration, impacted by generation profile, transmission congestion, and curtailment

The regions with the highest penetration of wind energy (SPP at 31%, ERCOT at 23%, and MISO at 11%) generally had the largest reduction to wind value, relative to the regional average value of a 24x7 flat-profile generator. The “value factor” of wind generation was roughly 0.5 in each of these high-penetration regions (value factor is calculated separately in each region and represents the ratio of the average value of wind generation to the average value of a 24x7 flat profile generation at all generator locations). Average wind value factors in the remaining regions were higher, equal to 0.8 in CAISO, ISO-NE and PJM, and 0.6 in NYISO. The progression of each region’s value factor with wind penetration can be seen in Figure 58. While there is a loose correlation between penetration level and value factor, each region’s value factor progressed along a convoluted path as penetration increased. Millstein et al. (2021) show that differences between the regions’ transmission infrastructure, and upgrades to that infrastructure, is one of the primary reasons value factor does not correlate more closely with penetration level.

55 Land-Based Wind Market Report

Wind Value Factor 1.2 CAISO ISO-NE 1.0

ERCOT 0.8 PJM 0.6

NYISO MISO SPP 0.4

0.2

0.0 0% 5% 10% 15% 20% 25% 30% 35% Wind Penetration Sources: Berkeley Lab, ABB, ISOs Figure 58. Trends in wind value factor as wind penetrations increase Using methods further described in Millstein et al. (2021), Figure 59 shows the impact of three separate causes of reduction to the value of wind generation in 2020. As used here, the term value reduction is the opposite of value factor: a total value reduction of 40% would indicate a value factor of 0.6. The three causes of value reduction are: (1) profile value reductions: caused by the temporal correlation of wind generation with low market prices, (2) congestion value reductions: caused by the inability to serve the most valuable locations in a region due to transmission congestion, and (3) curtailment value reductions: caused by curtailment of output, typically due to wind plant operator response to low (usually negative) local prices.

Figure 59 shows that the causes of wind value reductions vary from region to region. MISO, NYISO, and ERCOT face large congestion value reductions of 34%, 30%, and 27%. Of all the regions, SPP and ERCOT face the largest profile-based reductions of 31% and 18%. Curtailment value reductions did not reach above 2% in any region. These results suggest that new, intra-region transmission infrastructure might lead to the largest increases in wind value, relative to regional flat value, in ERCOT, MISO, and NYISO. Conversely, the large profile value reductions found in SPP, and to a lesser extent ERCOT, suggest that strategies beyond expansion of intra-regional transmission may be needed to enhance wind value. Millstein et al. (2021) discusses a range of possible strategies to address profile value reductions, from larger-scale inter-regional transmission and storage deployment, to new demand sources (e.g., coordinated electric vehicle charging) and regulatory and rate changes supporting shifting or more responsive load.

56 Land-Based Wind Market Report

ERCOT MISO SPP NYISO ISO-NE PJM CAISO Value of flat block $21/MWh $21/MWh $26/MWh $23/MWh $31/MWh $24/MWh $35/MWh

0%

$29/MWh -20% $25/MWh $20/MWh

-40% $13/MWh Profile Wind $11/MWh $14/MWh Congestion market $11/MWh value Curtailment -60%

Sources: Berkeley Lab, ABB, ISOs Figure 59. Impact of transmission congestion, output profile, and curtailment on wind energy market value in 2020

The health and climate benefits of wind are larger than its grid-system value, and the combination of all three far exceeds the levelized cost of wind

The benefits of wind in reducing health and climate burdens from polluting energy sources are not included in the earlier estimates of grid-system value and the comparisons of that value with PPA prices. Wind generation reduces power-sector emissions of carbon dioxide (CO2), nitrogen oxides (NOx), and sulfur dioxide (SO2). These reductions, in turn, provide public health and climate benefits (Millstein et al. 2017). In this section, the health and climate benefits of wind power are estimated and compared, along with grid-system value, to the unsubsidized levelized cost of new wind plants built in 2020.39

Using methods described in detail in the Appendix,40 Figure 60 presents the summed health and climate benefits from wind by region in 2020. Nationally, health and climate benefits together averaged $76/MWh- wind. Benefits were largest, ranging from $80/MWh to $120/MWh, in the Central, Midwest, and Mid-Atlantic regions (incorporating SPP, MISO, and PJM). Health and climate benefits were lowest in New York ($31/MWh) and New England ($27/MWh); and are not reported in the Southeast due to the small number of wind plants in that region. Regional and national values presented here include both in-region emission impacts as well as cross-region impacts due to electricity trade across regional boundaries.

39 The goal was to compare the most important cost and benefit components from a societal perspective, but this comparison is not exhaustive. Not included are considerations of employment; local environmental, ecological, land-use, and community impacts; water use; mercury and primary particulate ; and transmission or grid-integration costs not covered by grid-value estimates. 40 Briefly, the per-MWh health and climate benefits of wind were estimated through a two-step process: first, determine the marginal avoided emission rate; second, multiply avoided emissions by a regional damage rate (i.e., health or climate impacts per ton of pollutant emitted). Marginal avoided emission rates are derived from Fell and Johnson (2021). Damage rates for CO2 emissions are set to equal the social cost of carbon (IWG 2021; 2.5% discount rate), and health damage rates for SO2 and NOx come from EPA (2015). Health damage rates vary by the region in which the emissions occurred.

57 Land-Based Wind Market Report

Note: Estimates not provided for Southeast due to small number of wind plants in that region. Sources: Berkeley Lab, Form EIA-930, Fell and Johnson (2021) Figure 60. Marginal health and climate benefits from wind generation by region in 2020

Focusing on wind plants built in 2020, the average climate, health, and grid-system value sums to almost three times the average LCOE (see Figure 61). Specifically, climate, health, and grid-system values averaged $46/MWh, $31/MWh and $15/MWh, respectively, compared to an average LCOE of $33/MWh.

Costs and Benefits (2020 $/MWh)

100

80

Climate 60

40 Health 20 LCOE Grid 0 Cost Value Sources: Berkeley Lab, EIA Form 930, Fell and Johnson (2021)

Figure 61. Marginal health, climate and grid-value benefits from new wind plants versus LCOE in 2020

58 Land-Based Wind Market Report

For simplicity, single values for health and climate benefits are presented. However, these values represent central estimates from a range of plausible values. The central health values presented here are based on the average of high and low estimates, which are ±40% relative to the central value (representing a range of equally valid epidemiological research on the impact of human exposure to air pollutants). The climate benefits use a representative social cost of carbon from IWG (2021), but a range of estimates exist in the literature. Further discussion on the range of health impacts can be found in Millstein et al. (2017). Likewise, further discussion of the range of social cost of carbon estimates can be found in IWG (2021).

59 Land-Based Wind Market Report

Estimates of Wind Energy’s Integration Costs Wind energy output is variable, creating challenges for grid integration. Concerns about, and solutions to, these issues impact the pace of wind power deployment, and worldwide experience in operating power systems with wind energy highlights the critical role of power system flexibility. The figure below summarizes estimates of wind integration costs at various levels of wind penetration, from studies completed from 2003 through 2020, and grouped by region. While studies differ in how they define integration costs, the impacts assessed typically include any additional balancing costs associated with managing increased forecast errors and short-term variability. These costs are not included in the analysis in this chapter of the market value of wind, which only accounted for the hourly time-varying generation profile from wind and congestion facing wind in the system. Some of the integration cost studies reported in the figure also include estimates of the difference in value of wind with a time-varying profile compared to a more conventional dispatch profile, thereby potentially overlapping with the market value results presented in this chapter. The wind integration costs in these studies do not, however, include any costs associated with incremental transmission or the lower capacity contribution of wind, costs that are sometimes included in other integration cost estimates and that are partially captured in the market value estimates. Integration cost estimates are near or below $5/MWh in all of the regions shown, except the non-California portion of the Western Electricity Coordinating Council (WECC), for wind power capacity penetrations up to and even exceeding 60% of the peak load of the system in which the power is delivered. Studies in the non-California portion of WECC are all focused on individual utilities that also act as balancing authorities, with responsibility to maintain a balance between supply and demand at all times. These studies tend to find higher integration costs, though some recent studies find relatively low costs. Overall, the results of these studies show that costs tend to increase with wind penetration levels, and be lower when balancing areas are larger. Other variations in estimated costs are due, in part, to differences in methods, definitions of integration costs, power system and market characteristics, fuel price assumptions, wind forecasting details, and the degree to which thermal plant cycling costs are included.

Integration Cost ($/MWh) 12 CAISO ERCOT Eastern Interconnection MISO SPP 8 WECC (Non-CA) SERC

4

0 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% Wind Penetration (Capacity Basis) Notes: Green reflects ISO regions; pink non-ISO regions. All studies categorized as WECC (Non-CA) are from individual utilities within WECC. Studies in California and ERCOT are all regional. Many of the studies in the Eastern Interconnect (inclusive of those in MISO and SPP) are regional, but some are from individual utilities. Studies that assessed multiple wind penetrations using a common methodology are depicted with connecting lines. Sources: Citations for the studies included can be found in the data file, at: https://emp.lbl.gov/wind-technologies-market-report

60 Land-Based Wind Market Report

Transmission Investments and Wind Power

The areas with the greatest wind speeds are often distant from electricity load centers, making wind dependent on transmission infrastructure. Related, the low grid-system market value of wind in some areas of the country is, in part, driven by limited transmission and the resulting grid congestion.

New transmission additions were limited in 2020, with fewer than 1,000 miles of transmission lines coming online according to the Federal Energy Regulatory Commission (see figure below). The decline since the peak in 2013 is partly due to the completion of the transmission buildout in West Texas in 2013.

Completed Transmission (miles/year)

5,000 500 kV 345 kV 4,000 ≤ 230 kV

3,000

2,000

1,000

0 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Source: FERC monthly infrastructure reports

61 Land-Based Wind Market Report

9 Future Outlook Energy analysts project that annual wind additions will continue at a rapid clip in 2021, before declining for a couple years and then rebounding (BloombergNEF 2021a, Wood Mackenzie 2021b, GWEC 2021, EIA 2021c, IEA 2021). Among the forecasts for the domestic market presented in Figure 62, expected capacity additions range from 13 GW to 16 GW in 2021. Forecasts for 2022 and 2023 show a downturn. Annual additions then rebound to 11−13 GW in 2024 and 2025. These figures include both land-based and offshore wind.

These expected trends are driven in part by the scheduled expiration of the federal PTC, and by anticipated growth in offshore wind in the mid-2020s. Near-term additions are also influenced by improvements in the cost and performance of wind technologies (which contribute to low power sales prices), corporate wind energy purchases, and state-level renewable energy policies. Expectations for continued low natural gas prices put a damper on growth expectations, as do limited transmission infrastructure and competition from solar.

Annual Capacity (GW) 20

Analyst projections (bar = average) 15

10

5

0 2008 2018 1998 2000 2002 2004 2006 2010 2012 2014 2016 2020 2022e 2024e Sources: ACP, BloombergNEF (2021a), Wood Mackenzie (2021b), (GWEC 2021), (EIA 2021c), IEA (2021) Figure 62. Wind power capacity additions: historical installations and projected growth Longer term, the prospects for wind energy will be influenced by the sector’s ability to continue to improve its economic position even in the face of challenging competition from solar and natural gas. Corporate demand for clean energy and state-level policies will also continue to impact wind power deployment, as will the buildout of transmission infrastructure. Finally, the Biden Administration has established strong goals for clean energy, including a zero-carbon power sector by 2035. Consistent with those goals, there have been recent legislative proposals for a long-term extension of the PTC, a federal clean energy standard, and other national policies to support a clean energy transition. The fate of these legislative proposals will greatly impact the sector’s upside potential to exceed the projections shown above.

62 Land-Based Wind Market Report

References ABB. 2021. Suite Data Product. Accessed May 2021.

American Clean Power Association (ACP). 2021b. Clean Power Quarterly: 2021 Q1. Washington, D.C.: American Clean Power Association. https://cleanpower.org/resources/clean-power-report-q1-2021/

BloombergNEF (BNEF). 2017. Tax Equity Update: 2017. March 7.

BloombergNEF (BNEF). 2020. 2H 2020 Wind Turbine Price Index. December 2020.

BloombergNEF (BNEF). 2021a. 1H 2021 U.S. Renewable Energy Market Outlook. April 2021.

BloombergNEF (BNEF). 2021b. U.S. Trade Policy Cost Implications for Clean Energy. May 2021.

Energy Information Administration (EIA). 2021a. Annual Energy Outlook 2021. Washington D.C.: Energy Information Administration. https://www.eia.gov/outlooks/aeo/

Energy Information Administration (EIA). 2021b. Electric Power Monthly, with Data for February 2021. Washington D.C.: Energy Information Administration. https://www.eia.gov/electricity/monthly/

Energy Information Administration (EIA). 2021c. Short-Term Energy Outlook, May 2021. Washington D.C.: Energy Information Administration. https://www.eia.gov/outlooks/steo/

Federal Energy Regulatory Commission (FERC). 2021. Energy Infrastructure Update for January 2021 (and previous editions). Washington, D.C.: Federal Energy Regulatory Commission. https://cms.ferc.gov/media/energy-infrastructure-update-january-2021

Fell, H. and J.X. Johnson. 2021. “Regional disparities in emissions reduction and net trade from renewables.” Nature Sustainability, 4: 358-365.

Global Wind Energy Council (GWEC). 2021. Global Wind Report 2021. Brussels, Belgium: Global Wind Energy Council. https://gwec.net/global-wind-report-2021/

Hamilton, S., D. Millstein, M. Bolinger, and R. Wiser. 2020. “How Does Wind Project Performance Change with Age in the United States?” Joule 4: 1004-1020.

IEA. 2021. Renewable Energy Market Update: Outlook for 2021 and 2022. Paris, France: International Energy Agency. https://www.iea.org/reports/renewable-energy-market-update-2021

Interagency Working Group (IWG). 2021. Technical Support Document: Social Cost of Carbon, Methane, and Nitrous Oxide, Interim Estimates under Executive Order 13990. Washington D.C.: Interagency Working Group on Social Cost of Greenhouse . https://www.whitehouse.gov/wp- content/uploads/2021/02/TechnicalSupportDocument_SocialCostofCarbonMethaneNitrousOxide.pdf

LevelTen Energy. 2021. North America, Q2 2021: LevelTen Energy PPA Price Index. Seattle, Washington: LevelTen Energy. https://www.leveltenenergy.com/post/q2-2021

Millstein, D., R. Wiser, A.D. Mills, M. Bolinger, J. Seel, S. Jeong. 2021. “Solar and wind grid system value in the United States: The effect of transmission congestion, generation profiles, and curtailment.” Joule 5: 1-27.

Millstein, D., R. Wiser, M. Bolinger, and G. Barbose. 2017. “The climate and air-quality benefits of wind and solar power in the United States.” Nature Energy, 2: 17134.

63 Land-Based Wind Market Report

Moné, C., M. Hand, M. Bolinger, J. Rand, D. Heimiller, and J. Ho. 2017. 2015 Cost of Wind Energy Review. Golden, Colorado: National Renewable Energy Laboratory. https://www.nrel.gov/docs/fy17osti/66861.pdf

Norton Rose Fulbright. 2021. “Cost of Capital: 2021 Outlook.” Project Finance Newswire. February. https://www.projectfinance.law/publications/2021/february/cost-of-capital-2021-outlook/

Rand, J., M. Bolinger, R. Wiser and S. Jeong. 2021. Queued Up: Characteristics of Power Plants Seeking Transmission Interconnection As of the End of 2020. Berkeley, California: Lawrence Berkeley National Laboratory. https://emp.lbl.gov/publications/queued-characteristics-power-plants

United States Wind Turbine Database (USWTDB). U.S. Geological Survey, American Wind Energy Association, and Lawrence Berkeley National Laboratory data release. https://eerscmap.usgs.gov/uswtdb

U.S. Department of Energy (DOE). 2021. United States Energy & Employment Report 2021. Washington, DC: U.S. Department of Energy. https://www.energy.gov/us-energy-employment-jobs-report-useer

U.S. Environmental Protection Agency (EPA). 2015. Regulatory Impact Analysis for the Clean Power Plan Final Rule. Research Triangle Park, North Carolina: U.S. Environmental Protection Agency. https://www3.epa.gov/ttnecas1/docs/ria/utilities_ria_final-clean-power-plan-existing-units_2015-08.pdf

Wiser, R., M. Bolinger, and E. Lantz. 2019. “Assessing wind power operating costs in the United States: Results from a survey of wind industry experts.” Renewable Energy Focus 30: 46-57.

Wiser, R., and M. Bolinger. 2019. “Benchmarking Anticipated Wind Project Lifetimes: Results from a Survey of U.S. Wind Industry Professionals.” Berkeley, California: Lawrence Berkeley National Laboratory. https://eta-publications.lbl.gov/sites/default/files/wind_useful_life_report.pdf

Wood Mackenzie. 2021a. Global Wind Turbine Order Analysis: Q1 2021. March 2021.

Wood Mackenzie. 2021b. Global Wind Power Market Outlook Update: Q1 2021. March 2021.

64 Land-Based Wind Market Report

Appendix: Sources of Data Presented in this Report Installation Trends Data on wind power additions and repowering in the United States (as well as certain details on the underlying wind power projects) are sourced largely from ACP (2021a). Annual wind power capital investment estimates derive from multiplying wind power capacity data by weighted-average capital cost data (provided elsewhere in the report). Data on non-wind electric capacity additions come from ABB’s Velocity database, except that solar data come from Wood Mackenzie Power & Renewables.

Global cumulative (and 2020 annual) wind power capacity data are sourced from GWEC (2021) but are revised, as necessary, to include the U.S. wind power capacity used in the present report. Wind energy penetration is compiled by ACP (2021a).

The wind project installation map was created based on ACP’s project database. Wind energy as a percentage contribution to statewide electricity generation and consumption is based on EIA data for wind generation divided by in-state total electricity generation or consumption in 2020. Data on online hybrid power plants comes largely from EIA (updated, when erroneous data are discovered).

The wind hybrid/co-located data are compiled from the 2020 early release EIA 860 dataset. Projects are identified as hybrids with two approaches. The first approach involves identifying distinct power plants (e.g. wind and storage) that share the same EIA ID. This approach identifies the majority of the hybrid data summarized in the report. The second approach involves compiling data from ABB’s velocity suite and matching power plants that have the same ABB Plant ID but different fuel types. These plants were then found in the EIA dataset and cross-checked against and longitude information to confirm co-location.

Data on wind power capacity in various interconnection queues come from a review of publicly available data provided by each ISO or utility. For more information see Rand et al. (2021).

Industry Trends Turbine manufacturer market share data are derived from the ACP project database. Data on recent U.S. nacelle assembly capability come from ACP (2021a), as do data on U.S. tower and blade manufacturing capability. Manufacturer profitability data come from corporate financial reports.

Data on U.S. imports of selected wind turbine equipment come from the Department of Commerce, accessed through the U.S. Census Bureau, and obtained from the U.S. Census’s USA Trade Online data tool (https://usatrade.census.gov/). The analysis of the trade data relies on the “customs value” of imports as opposed to the “landed value” and hence does not include costs relating to shipping or duties. The table below lists the specific trade codes used in the analysis presented in this report.

All trade codes used to track wind equipment imported in 2020 are exclusive to wind. Prior to 2020, some codes are exclusive to wind, whereas others are not. Assumptions are made for the proportion of wind-related equipment in each of the non-wind-specific HTS trade categories. These assumptions are based on: an analysis of trade data where separate, wind-specific trade categories exist; a review of the countries of origin for the imports; personal communications with U.S. International Trade Commission and wind industry experts; U.S. International Trade Commission trade cases; and import patterns in the larger HTS trade categories.

65 Land-Based Wind Market Report

Table A1. Harmonized Tariff Schedule (HTS) Codes and Categories Used in Wind Import Analysis

HTS Code Description Years applicable Notes

includes both utility-scale and 8502.31.0000 wind-powered generating sets 2005–2020 small wind turbines not exclusive to wind turbine 7308.20.0000 towers and lattice masts 2006–2010 components 7308.20.0020 towers - tubular 2011–2020 mostly for wind turbines AC generators (alternators) from 750 to not exclusive to wind turbine 8501.64.0020 2006–2011 10,000 kVA components AC generators (alternators) from 750 to exclusive to wind turbine 8501.64.0021 2012–2020 10,000 kVA for wind-powered generating sets components not exclusive to wind turbine 8412.90.9080 other parts of and motors 2006–2011 components exclusive to wind turbine 8412.90.9081 wind turbine blades and hubs 2012–2020 components parts of generators (other than commutators, not exclusive to wind turbine 8503.00.9545 2006–2011 stators, and rotors) components parts of generators for wind-powered exclusive to wind turbine 8503.00.9546 2012–2020 generating sets components not exclusive to wind turbine machinery parts suitable for various machinery components; nacelles when 8503.00.9560 2014–2019 (including wind-powered generating sets) shipped without blades can be included in this category41 exclusive to wind turbine machinery parts for wind-powered generating components; nacelles when 8503.00.9570 2020 only sets shipped without blades are included in this category

Information on wind power financing trends was compiled by Berkeley Lab, based in part on data from the Intercontinental Exchange, BNEF, and Norton Rose Fulbright. Wind project ownership and power purchaser trends are based on a Berkeley Lab analysis of ACP’s project database. Technology Trends Information on turbine nameplate capacity, hub height, rotor diameter, and specific power was compiled by Berkeley Lab within the U.S. Wind Turbine Database based on information provided by ACP, turbine manufacturers, standard turbine specifications, the FAA, web searches, and other sources. The data include projects with turbines greater than or equal to 100 kW that began operation in 1998 through 2020. Estimates of the quality of the wind resource in which turbines are located were generated as discussed below.

FAA “Obstacle Evaluation / Airport Airspace Analysis” data containing prospective turbine locations and total proposed heights, in combination with ACP data on near-term installations, were used to estimate future technology trends. Any data with expiration dates between March 31, 2021 and September 30, 2022 were categorized as either “pending” turbines (for those that already had received an evaluation of “no hazard”) or “proposed” turbines (for those that were still being evaluated). A portion of those turbines are categorized by Berkeley Lab, with input from ACP data and ABB’s Velocity Suite data, as either “under construction” or in “advanced development.” The former are projects that have been partially or fully constructed but have not

41 The explicit inclusion of nacelles without blades was effective in 2014 as a result of Customs and Border Protection ruling number HQ H148455 (April 4, 2014). That ruling stated that nacelles alone do not constitute wind-powered generating sets, as they do not include blades—which are essential to wind-powered generating sets as defined in the HTS.

66 Land-Based Wind Market Report been fully commissioned. The latter are not under construction but are highly likely to be in the next few years and have one of the following in place: a signed PPA (or similar long-term contract), a firm turbine order, or an announcement to proceed under utility ownership.

Performance Trends Wind project performance data were compiled overwhelmingly from two main sources: FERC’s Electronic Quarterly Reports and EIA Form 923. Additional data come from FERC Form 1 filings and, in several instances, other sources. Where discrepancies exist among the data sources, those discrepancies are handled based on the judgment of Berkeley Lab staff. Data on curtailment are from ERCOT, MISO, PJM, NYISO, SPP, ISO-NE, and CAISO.

The following procedure was used to estimate the quality of the wind resource in which wind projects are (or are planned to be) located. First, within the U.S. Wind Turbine Database, the location of individual wind turbines and the year in which those turbines were (or are planned to be) installed were identified using FAA Digital Obstacle (i.e., obstruction) files and FAA Obstacle Evaluation / Airport Airspace Analysis files, combined with Berkeley Lab and ACP data on individual wind projects. Second, NREL used 200-meter resolution data from AWS Truepower—specifically, gross capacity factor estimates—to estimate the quality of the wind resource for each of those turbine locations. These gross capacity factors are derived from the average mapped 80-meter wind speed estimates, wind speed distribution estimates, and site elevation data, all of which are run through a standard wind turbine power curve (common to all sites) and assuming no losses. To create an index of wind resource quality, the resultant average wind resource quality (i.e., gross capacity factor) estimate for turbines installed in the 1998–1999 period is used as the benchmark, with an index value of 100%. Comparative percentage changes in average wind resource quality for turbines installed after 1998–1999 are calculated based on that 1998–1999 benchmark year. When segmenting wind resource quality into categories, the following AWS Truepower gross capacity factors are used: the “lower” category, which includes all projects or turbines with an estimated gross capacity factor of less than 40%; the “medium” category, which corresponds to ≥40%–45%; the “higher” category, which corresponds to ≥45%–50%; and the “highest” category, which corresponds to ≥50%. Not all turbines could be mapped by Berkeley Lab for this purpose; the final sample included 61,579 turbines of the 62,324 installed from 1998 through 2020 in the continental United States (i.e., nearly 99%). Most of the turbines that are not mapped are more than a decade old.

Separate from the above, the relative strength of the average “fleet-wide” wind resource from year to year is estimated based on weighting each operational project-level wind resource (or “wind index”) by its share of the total operational fleet-wide capacity for the particular year. For each individual wind plant, an annual wind index is calculated as the ratio of a particular year’s predicted capacity factor to the long-term average predicted capacity factor (with the long-term average calculated from 1998-2020). Site-level available wind resources are calculated for each hour of each year based on ERA5 reanalysis wind speed data for each plant’s location. ERA5 has a horizontal resolution of ~30 km × 30 km. Site-specific estimated wind speeds (with the geographic resolution previously noted) are interpolated between ERA5 model heights to the corresponding representative hub-height for each wind project. Hourly wind speeds at each project are then converted to wind power by applying project-specific power curves. In this case, power curves are based on the set of turbine- specific power curves reported by thewindpower.net, which provides power curves for more than 750 separate turbines; some newer power curves are derived from NREL’s System Advisor Model, v2020.11.29 and based on turbine characteristics, such as specific power. Although many projects contain only a single type of turbine, some projects contain multiple turbine types. For the latter projects, a turbine power curve is selected that most closely matches the average turbine capacity, rotor diameter, and specific power across the project. The wind indices are calculated without accounting for wake, electrical, or other losses, or curtailment, and are based only on the ERA5 wind speeds. These indices are used to represent changes in the wind resource from one year to the next, and reflect the ERA5-based strength of the total potential wind resource given the types of turbines that are deployed at each site. Note that these data and indices are used to characterize year-to-year variations in the strength of the wind resource, whereas AWS Truepower estimates are used to characterize the

67 Land-Based Wind Market Report strength of the site-specific long-term annual average wind resource. The analyses uses AWS Truepower estimates for the latter need due to their higher geographic resolution.

Cost Trends Historical U.S. wind turbine transaction prices were, in part, compiled by Berkeley Lab. Sources of transaction price data vary, but most derive from press releases, press reports, and Securities and Exchange Commission and other regulatory filings. Additional data come from Vestas, SGRE and Nordex corporate reports, BloombergNEF, and Wood Mackenzie.

Berkeley Lab used a variety of public and some private sources of data to compile capital cost data for a large number of U.S. wind projects. Data sources range from pre-installation corporate press releases to verified post-construction cost data. Specific sources of data include EIA Form 412, EIA Form 860, FERC Form 1, various Securities and Exchange Commission filings, filings with state public utilities commissions, Windpower Monthly magazine, AWEA’s Wind Energy Weekly, the DOE and Electric Power Research Institute Turbine Verification Program, Project Finance magazine, various analytic case studies, and general web searches for news stories, presentations, or information from project developers. For 2009–2012 projects, data from the Section 1603 Treasury Grant program were used extensively; for projects installed from 2013 through 2019, EIA Form 860 data are used extensively. Some data points are suppressed in the figures to protect data confidentiality. Because the data sources are not all equally credible, less emphasis should be placed on individual project-level data; instead, the trends in those underlying data offer greater insight. Only cost data from the contiguous lower-48 states are included.

Wind project O&M costs come primarily from two sources: EIA Form 412 data from 2001 to 2003 for private power projects and projects owned by POUs, and FERC Form 1 data for IOU-owned projects. A small number of data points are suppressed in the figures to protect data confidentiality. Sales Price and Levelized Cost Trends Wind PPA price data are based on multiple sources, including prices reported in FERC’s Electronic Quarterly Reports, FERC Form 1, avoided-cost data filed by utilities, pre-offering research conducted by bond rating agencies, and a Berkeley Lab collection of PPAs. Supplemental data from Level10 Energy are also reported. REC price data were compiled by Berkeley Lab based on information provided by Marex Spectron.

The analysis calculates the LCOE of wind based on LCOE input data collected, in large part, by Berkeley Lab and presented elsewhere in this report. These inputs include capital costs, capacity factors, operational expenses, financing costs, and assumptions about useful life. Specifically: • For capacity factors, project-level data are levelized over the assumed useful life of each plant, applying degradation assumptions from Hamilton et al. (2020) as appropriate. For projects built in 2020 (that have not yet been operating for a full year), capacity factors are assumed to match the average capacity factor of projects built in the same regions from 2016 to 2018. • Based on Wiser et al. (2019), total operational expenses are assumed to fall from a levelized cost of $84/kW-year in 1998 to $63/kW-year by 2003, $54/kW-year by 2010, and $45/kW-year by 2018 (and are interpolated linearly between these years). Projects built in 2019 and 2020 assume that levelized operational expenses have continued to decline at the same rate as from 2017 to 2018. • The weighted average cost of capital assumes a 70%:30% debt-to-equity split (possible in the absence of the PTC), with the cost of debt varying over time based on historical changes in the 20- and 30-year swap rates and bank spread, while the cost of equity declines from 15% in 1998 to 8.25% in 2020. Financing costs are estimated as if the PTC were not available. • Project life is assumed to increase linearly from 20 years for projects built in 1998 to 30 years for projects built in 2020 (see Wiser and Bolinger 2019).

68 Land-Based Wind Market Report

• A 35% corporate tax rate is assumed from 1998–2017 and 21% thereafter, with a constant 5% state tax rate over the entire period. Inflation expectations range from 1.9% to 3.0%. Five-year accelerated depreciation is applied for all vintages of wind projects.

Cost and Value Comparisons To compare the price of wind to the cost of future natural gas-fired generation, the range of fuel cost projections from the EIA’s Annual Energy Outlook 2021 is converted from $/MMBtu into $/MWh using heat rates derived from the modeling output.

To calculate the historical wholesale energy market value of wind, estimated hourly wind generation profiles are matched to hourly nodal real-time wholesale prices. The capacity value at each plant is also calculated, based on the modeled wind profiles and ISO-specific rules for wind’s capacity credit and ISO-zone-specific capacity prices. The resulting estimates reflect the average $/MWh energy and capacity value for each plant and year. ISO-level average values are estimated by weighting plant-level value estimates by plant output.

To calculate the average energy and capacity value in $/MWh, the numerator is based on actual hourly generation after curtailment but the denominator is based on the total generation without curtailment. Curtailment is accounted for only in the numerator so that increased levels of curtailment will reduce the average $/MWh value. The MWh, in this case, reflect potential wind generation before curtailment. Note that public data do not broadly exist for hourly wind output profiles at the plant level. Consequently, the ERA5- based modeled wind generation estimates described earlier are leveraged, albeit adjusted for curtailment and corrected for bias. The resulting generation estimates incorporate publicly available information on actual generation as well as site-specific ERA5 modeled wind speeds. One exception to this process is for plants located in ERCOT. ERCOT provided high time-resolution records of plant level generation and curtailment going back to 2013, and, where available, those reported values are utilized.

Total curtailment is reported by each ISO for either each hour or each month. To correct ERA5 output estimates for curtailment, plants are divided into three groups: plants receiving the PTC, plants that have aged out of the PTC, and plants that elected the 1603 Treasury Grant instead of the PTC. Total reported hourly curtailment is distributed evenly across all plants within a particular ISO that face local hourly prices below a threshold defined for each group (initially, –$23/MWh for PTC plants and $0/MWh for the other two groups). A similar process is used to distribute monthly curtailment data. Bias correction involves an iterative linear scaling approach so that: (1) each plant’s total modeled generation matches its reported generation (from EIA or FERC, typically at the monthly or quarterly level), and (2) the sum of estimated hourly generation across all plants within each ISO matches the hourly total wind generation reported by each ISO.

Hourly nodal real-time wholesale electricity prices and hourly regional wind output profiles are from ABB’s Velocity Suite database. Curtailment data are downloaded directly from each ISO, or in some cases, from ABB’s Velocity Suite database. For each wind power plant, the nearest or most-representative pricing node is identified, which allows representative prices to be matched to each plant. For some regions, hourly wind output profiles are only available for a subset of the relevant years of the analysis; as such, estimates of the wholesale energy value of wind are not available for all years for all regions.

Capacity value is estimated for each plant based on the bias-corrected, modeled wind profiles and ISO and ISO-zone specific capacity prices or costs, as well as relevant regional rules for wind’s capacity credit. A separate capacity value is not calculated for ERCOT, because ERCOT runs an energy-only market that does not require load-serving entities to meet a resource adequacy obligation. In ERCOT, however, hourly Operating Reserve Demand Curve prices are added to nodal energy prices. Capacity value in ERCOT is essentially incorporated into the energy markets. As for capacity prices and costs, many regions have organized capacity markets. In those cases, the analysis uses market-clearing prices from capacity market auctions in concert with ISO-rules or estimates for the capacity credit of wind. For regions where load-serving

69 Land-Based Wind Market Report entities have a resource adequacy obligation but lack organized capacity markets, the analysis uses data from regulatory bodies to approximate capacity costs and regional estimates or rules for wind’s capacity credit.

The analysis calculates the difference between wind value and flat-profile value (called “value reduction”) and then further decomposes the value reduction into three separate causes: profile, congestion, and curtailment. Flat profile value is calculated in two steps. First, the average value of flat (“always-on”) generation is calculated at all power plant pricing nodes in a region (both wind and other types of power plants). The regional flat value is then calculated by taking the weighted-average value across all these power plants with weights based on recorded energy output at each plant. The profile value of wind is calculated in a similar manner to the regional flat value, but instead of using a flat profile, a wind plant output profile is applied to all power plants in a region (both wind and other types) and the regional weighted average value is calculated. This process is repeated for the profiles for all wind plants in a region to develop the regional average wind plant profile value. The reduction in wind value due to its profile is then calculated as the difference between the regional wind profile value and the regional flat value. Next, the value of wind generation at each wind plant is calculated given its output profile, and the regional average value is calculated across all wind plants. This provides a value of wind profiles at wind plants—in effect, the value of wind generation (not yet adjusted for curtailment). The profile value calculation finds the value of wind output at all generator locations and the wind generation value finds wind value only at wind generators, so the difference represents the impact of transmission congestion. Finally, the value of wind is adjusted for curtailment by increasing the total energy over which energy and capacity revenue are normalized. This final adjustment provides the overall value of wind at each plant. These methods are described in further detail in Millstein et al. (2021).

Turning to health and climate benefits, the marginal rate of health benefits is estimated based on a two-step process. First, the marginal rate of avoided emissions for wind is calculated based on estimates developed by Fell and Johnson (2021) for nine regions of the United States. Fell and Johnson (2021) also estimate the impact of wind on emissions in neighboring regions, and these additional impacts are included in the present analysis. An exception is that Fell and Johnson (2021) do not estimate the emission impacts on neighboring regions for wind generation within New York, New England, and Texas. Fell and Johnson (2021) do, however, estimate wind’s per MWh impact on net exports for all regions. This impact on net exports is used in the present analysis to estimate the cross region impact on emissions where Fell and Johnson (2021) do not calculate it directly. The exported wind energy is assumed here to cause a reduction to natural gas generation, with natural gas emission rates by region calculated based on the EPA’s eGRID2019 data.

Note that the Fell and Johnson (2021) estimates are based on regressions that used data over the period July 2018 through March 2020; as such, the emission factors used here do not precisely reflect calendar year 2020. As well, the Fell and Johnson (2021) regressions find that an increase in wind offsets a small amount of hydropower. Over the course of a year, however, hydropower generation can be assumed to be fixed. To estimate the emissions impacts of wind in the current analysis, wind is assumed to shift hydropower generation in time, which in turn reduces another source of generation. The type of generation reduced in response to the shifting of hydropower is unknown; to maintain a conservative estimate, natural gas is assumed to be reduced, again employing regional emissions rates from eGRID2019.

A reduced-order health impacts model is then used to estimate the value of the avoided emissions from wind. Reduced-order health impacts models use the results of full meteorological and air quality models to provide more generalized estimates of the marginal impacts of emissions from specific regions. This analyses uses a model developed in EPA (2015), which contains marginal impact estimates (as dollars of health damage per ton of emitted SO2 and NOx emissions) for power-sector emissions in three large regions for 2020. EPA (2015) provides a high and low estimate for the marginal damage rate, based on differing epidemiological studies. This analysis uses marginal damage estimates from the EPA based on a 3% discount rate. The product of these damage estimates with the marginal emission rate provides a monetized marginal benefit per MWh of wind generation. The estimated health benefits include reduced hospitalizations and reduced work-days missed, but the monetization is dominated by the cost of premature mortality due to population exposure to .

70 Land-Based Wind Market Report

The value of avoided CO2 emissions due to wind was calculated in a similar manner. Specifically, marginal CO2 emissions factors were also derived from Fell and Johnson (2021), and include the additional processing described above to account for hydropower impacts in all regions and exports in New York, New England, and Texas. The marginal emission factors were then multiplied by the social cost of carbon from IWG (2021), using the 2.5% discount rate case, to derive a monetized per-MWh benefit for wind generation by region.

Both estimates of health and climate benefits are subject to uncertainty. Central estimates of these benefits are presented, though both estimates have a wide range of plausible values. More detail about this uncertainty is available in Millstein et al. (2017) and Fell and Johnson (2021).

71 Land-Based Wind Market Report: 2021 Edition

For more information visit,  energy.gov/eere/wind

DOE/GO-102021-5611 • August 2021

Cover details: near Macksburg, Iowa. Photo courtesy of Ed Collier / Ed Collier Photography.